responses feedback items: whitelist input-shape fields — api rejects response-only fields like status as unknown parameters (live 400)

DECISIONS.md

@@ -60,6 +60,47 @@ Decisions inside the set architecture. D-NNN, never renumbered.
   broken classifier must never mute the bot; the budget gate already
   bounds spend. Its verdict gates BEFORE the main call, the
   envelope's answer_needed still gates after — two independent nets.
+- **D-021** — Health monitoring (FDB-012, SPEC-012 OPS-18/19): a
+  separate `monitor_loop` (own cadence, default 300 s) rather than
+  folding checks into the 60 s task loop — monitoring is coarse and
+  should not run every minute. Checks are edge-triggered (alert on the
+  rising edge, re-arm on recovery) so a standing condition never spams;
+  they reuse the existing rate-limited staff-alert path. Metrics are
+  the cheap, high-signal ones (spend vs budget, free disk, task-queue
+  depth); each is independently skippable when it has no data, so a
+  deployment without a budget or store still runs the others. Opt-in
+  (`enable-monitoring`) like every other operational rollout.
+- **D-021** — Responses API behind `use-responses-api` (FDB-028,
+  ENV-22..24, resolves D-006): the responder path can use
+  `/v1/responses`, which allows tools + `reasoning_effort` (the
+  chat/completions 400 from ENV-21) and keeps one chain of thought
+  across tool rounds. Stateless by choice: `store=false` +
+  encrypted reasoning items passed back — GDPR posture unchanged, no
+  server-side conversation retention. Flag defaults off; rollback is
+  a config toggle (hot-reload), not a deploy. Classifier /
+  consolidation / task-gen stay on chat/completions (no tools, no
+  reasoning need — not worth the churn).
+- **D-020** — Web search via Exa (FDB-022, SPEC-015): a `web_search`
+  tool alongside fetch_url/IGDB/codex/get_news, filling the "look it up
+  on the open web" gap. Exa (not a raw search-engine scrape) because it
+  returns clean title+url+text in one call — no SSRF surface of our own
+  (we call one fixed API endpoint, not arbitrary hosts), and it pairs
+  with fetch_url for the full article. Key is a host secret
+  (`exa-api-key`, env `EXA_API_KEY` fallback), never repo-side; results
+  sanitized like every other external-text tool; off by default
+  (`enable-web-search`), metered per user.
+- **D-019** — News memory + on-demand tool (SPEC-013 NEWS-07..12):
+  the news pipeline now carries item summaries (feed descriptions,
+  HTML-stripped) and persists every fetched item into a deduped `news`
+  table (schema v6), pruned to a rolling window (`news-keep`). Both the
+  kroa digest run and the ggg posting run write to it, so the store is
+  a single searchable source across both models. A `get_news` tool
+  reads that store (topic/source-filtered, metered, sanitized) rather
+  than re-fetching feeds live: the ambient `{news}` digest stays a
+  small always-on snapshot, while the tool gives unbounded on-demand
+  reach without a fresh network round-trip per call. The store is the
+  same `bot.db` (WAL) the bot uses; the cron process opens it
+  independently — concurrent reader/writer is what WAL is for.
 - **D-018** — Codex Mechanicus search (FDB-019, SPEC-014): Luma's
   lore is grounded in the priest's real archive at binaric.tech via a
   `codex_search` tool over the site's public `search-index.json`, not

Makefile

@@ -1,6 +1,6 @@
 # Fjerkroa Bot Development Makefile (uv-managed)
 
-.PHONY: help install install-dev clean test test-cov test-fast lint format format-check type-check security-check audit trace check all-checks pre-commit run run-dev build ci
+.PHONY: deploy backup help install install-dev clean test test-cov test-fast lint format format-check type-check security-check audit trace check all-checks pre-commit run run-dev build ci
 
 # Default target
 help: ## Show this help message

docs/RUNBOOK.md

@@ -0,0 +1,130 @@
+# Operator runbook — Fjærkroa / Luma bot
+
+One page for "something is wrong, what do I do". Two deployments of one
+codebase, both on **uberspace** (push-based deploy from the dev machine —
+there is no git checkout on the hosts).
+
+| | Fjærkroa (café) | Luma (GGG clan) |
+| --- | --- | --- |
+| SSH host | `ssh fjerkroa` (pictor.uberspace.de) | `ssh ggg` |
+| Service | `kroa` | `luma` |
+| Config | `~/fjerkroa_bot/kroa.toml` | `~/fjerkroa_bot/ggg.toml` |
+| Staff channel | `#kassa` | `#mods` |
+| Language / persona | Norwegian, café host | German, "Luma" |
+
+Common paths on each host: bot code `~/fjerkroa_bot`, venv `~/venv-bot`,
+database `~/fjerkroa_bot/history/bot.db` (SQLite, WAL), our snapshots
+`~/backups/<kroa|luma>/`, logs under `~/logs` and `~/tmp`.
+
+## From Discord (staff channel only, prefix `!bot`)
+
+No SSH needed for day-to-day control. Type `!bot help` in the staff
+channel for the full, grouped list. The essentials:
+
+- `!bot pause` / `!bot resume` — stop / start all replies.
+- `!bot quiet <minutes>` — go silent for a while, then auto-resume.
+- `!bot status` — replies/images/tasks flags + quiet time left.
+- `!bot spend` — today's estimated USD spend, tokens, images, budget.
+- `!bot images on|off`, `!bot tasks on|off` — kill-switches.
+
+`!help` works in **any** channel (for everyone) and lists only what is
+usable there. `!forgetme` and `!privacy` also work everywhere, even
+while the bot is paused.
+
+## Restart / check health (SSH)
+
+```sh
+ssh <host>
+supervisorctl status <kroa|luma>          # RUNNING + uptime
+supervisorctl restart <kroa|luma>
+tail -n 40 ~/tmp/<kroa|luma>-stderr*.log   # discord login / errors
+tail -n 40 ~/logs/supervisord.log          # "We have logged in as ..."
+```
+
+A healthy start shows a fresh `connected to Gateway` + `We have logged
+in as ...` line within ~15 s.
+
+## Deploy a release / roll back
+
+From the **dev machine** (`~/Repos/FjerkroaBot`), tags only:
+
+```sh
+git tag -m "<msg>" vX.Y.Z && git push --tags        # cut the release first
+bash deploy/deploy.sh ggg vX.Y.Z                    # luma
+DEPLOY_FORCE=1 bash deploy/deploy.sh fjerkroa vX.Y.Z # kroa (see window)
+```
+
+- kroa refuses to deploy **11:00–22:00 Europe/Oslo** (restaurant hours);
+  `DEPLOY_FORCE=1` overrides. Café is closed Mondays.
+- The script backs up `bot.db` → `bot.db.pre-<tag>` before restart, then
+  smoke-tests (RUNNING + fresh login) and fails loudly if either misses.
+- **Rollback** = deploy the previous tag. If the schema version moved
+  between the two tags, restore the matching `bot.db.pre-<newtag>` first
+  (see below) so the older code meets a schema it understands.
+
+## Restore the database
+
+Three independent daily backup layers exist — pick the freshest good one.
+
+```sh
+ssh <host>
+supervisorctl stop <kroa|luma>
+DB=~/fjerkroa_bot/history/bot.db
+
+# 1) uberspace nightly backup of the whole home (read-only):
+#    /backup = current + daily.0..7 + weekly.1..7  (15 restore points)
+cp /backup/daily.1/home/<user>/fjerkroa_bot/history/bot.db "$DB"
+
+# 2) our own rotated gzip snapshot (03:17 UTC cron, keep 14):
+gunzip -c ~/backups/<kroa|luma>/bot-YYYYMMDD-HHMMSS.db.gz > "$DB"
+
+# 3) the pre-deploy snapshot for a given release:
+cp "$DB".pre-vX.Y.Z "$DB"
+
+rm -f "$DB"-wal "$DB"-shm    # drop stale WAL sidecars after a restore
+supervisorctl start <kroa|luma>
+```
+
+`<user>` is `fjerkroa` or `ggg`. The DB holds conversation history,
+structured memory, usage ledger, image cache index, tasks, and the news
+store — all regenerable, none critical. That is why there is no off-host
+backup: uberspace `/backup` + the on-host snapshots are enough.
+
+## Rotate a secret
+
+Secrets live only in the host `*.toml` (never in the repo). Edit in
+place and restart:
+
+```sh
+ssh <host>
+# OpenAI: edit  openai-token = "sk-..."   in kroa.toml / ggg.toml
+# Discord: edit discord-token = "..."     (get a new token from the
+#          Discord developer portal → Bot → Reset Token first)
+supervisorctl restart <kroa|luma>
+```
+
+After rotating an OpenAI key, revoke the old one in the OpenAI dashboard.
+Keep a `*.toml` backup before editing; a broken TOML crash-loops the
+service (validate: `~/venv-bot/bin/python -c 'import tomlkit; tomlkit.load(open("kroa.toml"))'`).
+
+## Scheduled jobs (crontab -l)
+
+| Host | When (server time) | Job |
+| --- | --- | --- |
+| both | `17 3 * * *` | `backup_db.py` → `~/backups/<bot>/` (keep 14) |
+| kroa | `5 * * * *` | news digest → `{news}` file + news store |
+| ggg | `*/15 * * * *` | news poster → #news/#newsjp webhooks + store |
+
+Logs: `~/backups/<bot>/backup.log`, `~/backups/<bot>/news*.log`.
+
+## Quick triage
+
+- **Bot silent everywhere** → `!bot status` (paused/quiet?), else
+  `supervisorctl status`; if not RUNNING, `restart` and read stderr.
+- **Bot silent in one channel** → check the host config `ignore-channels`
+  / `short-path` rules for that channel (a stray `short-path` rule can
+  archive messages without replying).
+- **Repeated API errors** → the bot posts a rate-limited alert to the
+  staff channel after 5 consecutive OpenAI failures (OPS-16); check
+  `!bot spend` (budget hit?) and the OpenAI status/key.
+- **Bad deploy** → roll back to the previous tag (above).

fjerkroa_bot/ai_responder.py

@@ -105,6 +105,7 @@ class AIMessage(AIMessageBase):
         self.channel = channel
         self.direct = direct
         self.historise_question = historise_question
+        self.factual = False  # classifier verdict; may route to factual-model (BEH-10)
         self.vars = ["user", "message", "channel", "direct", "historise_question"]
 
 
@@ -340,6 +341,9 @@ class AIResponder(AIResponderBase):
         # Get the history limit from the configuration
         limit = self.config["history-limit"]
 
+        # Factual verdict routes this call to factual-model if configured (BEH-10)
+        self._factual = bool(getattr(message, "factual", False))
+
         # Check if a short path applies, return an empty AIResponse if it does
         if self.short_path(message, limit):
             await self._persist_history()

fjerkroa_bot/codex.py

@@ -1,7 +1,7 @@
 """Codex Mechanicus search tool (SPEC-014, FDB-019).
 
 Luma's own sacred archive — the Codex Mechanicus at binaric.tech — as a
-function tool. She searches the codex index and answers Cult Mechanicus
+function tool. He searches the codex index and answers Cult Mechanicus
 lore from real, sourced inscriptions instead of inventing it. The index
 is fetched over HTTPS (SSRF-guarded, size-bounded, cached in memory) and
 every field returned to the model is sanitized (SAF-03), because even

fjerkroa_bot/discord_bot.py

@@ -1,11 +1,14 @@
 import argparse
 import asyncio
+import fnmatch
 import logging
 import random
 import re
+import shutil
 import sys
 import time
 from collections import deque
+from pathlib import Path
 from typing import Optional, Union
 
 import discord
@@ -16,6 +19,7 @@ from watchdog.events import FileSystemEventHandler
 from watchdog.observers import Observer
 
 from .ai_responder import AIMessage
+from .monitor import HealthMonitor
 from .openai_responder import OpenAIResponder
 from .tasks import TaskEngine
 
@@ -65,11 +69,41 @@ def split_answer(text: str, threshold: int, max_parts: int) -> list:
 
 
 class ConfigFileHandler(FileSystemEventHandler):
-    def __init__(self, on_modified):
-        self._on_modified = on_modified
+    """Rename-safe config watch (CFG-05).
 
+    Editors and tools save atomically — write a temp file, then rename it
+    over the target — which fires a *moved*/*created* event (not
+    *modified*) and swaps the inode, so watching the file directly goes
+    deaf after the first save. We watch the config's *directory* and react
+    to any event whose src or dest path is the config file.
+    """
+
+    def __init__(self, config_path: str, on_change):
+        self._config_path = str(Path(config_path).resolve())
+        self._on_change = on_change
+
+    def _hits_config(self, event) -> bool:
+        for attr in ("src_path", "dest_path"):
+            path = getattr(event, attr, "")
+            if path and str(Path(path).resolve()) == self._config_path:
+                return True
+        return False
+
+    def _dispatch(self, event):
+        if not event.is_directory and self._hits_config(event):
+            self._on_change()
+
+    # Only write/rename events — NOT on_opened/on_closed, whose read-opens
+    # (our own load_config re-reads the file) would otherwise feed back into
+    # a reload loop (CFG-05).
     def on_modified(self, event):
-        self._on_modified(event)
+        self._dispatch(event)
+
+    def on_created(self, event):
+        self._dispatch(event)
+
+    def on_moved(self, event):
+        self._dispatch(event)
 
 
 class FjerkroaBot(commands.Bot):
@@ -99,8 +133,10 @@ class FjerkroaBot(commands.Bot):
 
     def init_observer(self):
         self.observer = Observer()
-        self.file_handler = ConfigFileHandler(self.on_config_file_modified)
-        self.observer.schedule(self.file_handler, path=self.config_file, recursive=False)
+        config_path = Path(self.config_file).resolve()
+        self.file_handler = ConfigFileHandler(str(config_path), self.on_config_file_changed)
+        # Watch the directory, not the file — atomic saves replace the inode (CFG-05)
+        self.observer.schedule(self.file_handler, path=str(config_path.parent), recursive=False)
         self.observer.start()
 
     def init_aichannels(self):
@@ -130,6 +166,15 @@ class FjerkroaBot(commands.Bot):
             observe=self.airesponder.observe_event,
         )
         self.loop.create_task(self.task_loop())
+        # Proactive health monitoring -> staff alerts (OPS-18/19)
+        self.health_monitor = HealthMonitor(
+            config_getter=lambda: self.config,
+            ledger=self.airesponder.ledger,
+            store=self.airesponder.store,
+            disk_free_mb=self._disk_free_mb,
+            alert=self.send_staff_alert,
+        )
+        self.loop.create_task(self.monitor_loop())
         logging.info("Task engine initialised.")
 
     async def task_loop(self):
@@ -140,6 +185,20 @@ class FjerkroaBot(commands.Bot):
             except Exception as err:
                 logging.warning(f"task tick failed: {repr(err)}")
 
+    def _disk_free_mb(self) -> float:
+        directory = Path(self.config.get("history-directory", ".")).expanduser()
+        target = directory if directory.exists() else Path.home()
+        return shutil.disk_usage(target).free / (1024 * 1024)
+
+    async def monitor_loop(self):
+        while True:
+            await asyncio.sleep(int(self.config.get("monitor-interval", 300)))
+            if self.health_monitor.enabled():
+                try:
+                    await self.health_monitor.tick()
+                except Exception as err:
+                    logging.warning(f"monitor tick failed: {repr(err)}")
+
     async def _execute_task(self, channel_name: str, prompt: str) -> None:
         """Run a due task through the normal responder path (TSK-02)."""
         channel = self.channel_by_name(channel_name, getattr(self, "chat_channel", None), no_ignore=True)
@@ -182,6 +241,9 @@ class FjerkroaBot(commands.Bot):
         if content.startswith("!privacy"):
             await message.channel.send(self.config.get("privacy-notice", DEFAULT_PRIVACY_NOTICE), suppress_embeds=True)
             return
+        if content.startswith("!help"):  # OPS-17: context-aware, works even while paused
+            await message.channel.send(self._help_text(staff=self.is_staff_channel(message.channel)), suppress_embeds=True)
+            return
         if not self.replies_allowed():
             return
         if str(message.content).startswith("!wichtel"):
@@ -263,15 +325,35 @@ class FjerkroaBot(commands.Bot):
             return f"Cancelled {store.task_set_state(int(args[1]), 'cancelled')} task(s)."
         return None
 
+    def _help_text(self, staff: bool) -> str:
+        """Context-aware command help (OPS-17): every channel lists the user commands; the staff channel also lists operator commands."""
+        everywhere = (
+            "Available to everyone, in any channel:\n"
+            "• `!help` — this help\n"
+            "• `!forgetme` — delete your messages and memory traces (works even while I'm paused)\n"
+            "• `!privacy` — how your data is handled (works even while I'm paused)\n"
+            "• `!wichtel @a @b @c …` — draw Secret Santa pairings (needs ≥2 mentions; only while I'm active)"
+        )
+        if not staff:
+            return everywhere
+        operator = (
+            "Staff commands — this channel only, prefixed `!bot`:\n"
+            "• Control: `pause`, `resume`, `quiet <minutes>`, `status`\n"
+            "• Cost: `spend`, `images on|off`\n"
+            "• Memory: `memory <user>`, `forget-fact <id>`, `pin <channel|global> <text>`, `unpin <id>`, `pins`\n"
+            "• Tasks: `tasks` (list), `tasks on|off`, `task-approve <id>`, `task-cancel <id>`"
+        )
+        return operator + "\n\n" + everywhere
+
     async def handle_staff_command(self, message: Message) -> None:
-        """Operator kill-switches, staff channel only (OPS-01..05, OPS-09, MEM-07)."""
+        """Operator kill-switches, staff channel only (OPS-01..05, OPS-09, OPS-17, MEM-07)."""
         args = str(message.content).split()[1:]
         for handler in (self._memory_command, self._task_command):
             reply = handler(args)
             if reply is not None:
                 await message.channel.send(reply, suppress_embeds=True)
                 return
-        reply = "Commands: pause, resume, images on|off, tasks on|off, quiet <minutes>, status, spend, memory <user>, forget-fact <id>, pin <channel|global> <fact>, unpin <id>"
+        reply = self._help_text(staff=True)  # OPS-17: unknown/`help` -> full grouped help
         if args[:1] == ["pause"]:
             self.replies_enabled = False
             reply = "Replies paused."
@@ -366,12 +448,10 @@ class FjerkroaBot(commands.Bot):
             airesponder.image_cache.purge_message(str(message.id))  # IMG-14
         await airesponder.observe_event(message.author.name, "delete", f"deleted: {message.content}")
 
-    def on_config_file_modified(self, event):
+    def on_config_file_changed(self):
         # Runs on the watchdog observer thread — the swap itself is
         # scheduled onto the event loop so no request reads a
         # half-swapped config (CFG-04 / D9)
-        if event.src_path != self.config_file:
-            return
         new_config = self.load_config(self.config_file)
         if repr(new_config) == repr(self.config):
             return
@@ -402,7 +482,7 @@ class FjerkroaBot(commands.Bot):
             return fallback_channel
         if channel_name.startswith("#"):
             channel_name = channel_name[1:]
-        if not no_ignore and channel_name in self.config.get("ignore-channels", []):
+        if not no_ignore and self.channel_ignored(channel_name):
             return fallback_channel
         for guild in self.guilds:
             channel = discord.utils.get(guild.channels, name=channel_name)
@@ -415,8 +495,12 @@ class FjerkroaBot(commands.Bot):
             return str(channel.recipient.name)
         return str(channel.id) if isinstance(channel, DMChannel) else str(channel.name)
 
+    def channel_ignored(self, channel_name) -> bool:
+        """fnmatch patterns; plain names match exactly as before (BEH-09)."""
+        return any(fnmatch.fnmatchcase(str(channel_name), pattern) for pattern in self.config.get("ignore-channels", []))
+
     def ignore_message(self, channel_name, message):
-        return channel_name in self.config.get("ignore-channels", []) and not message.direct
+        return self.channel_ignored(channel_name) and not message.direct
 
     def log_message_action(self, action, message, channel_name):
         logging.info(f"{action} message {repr(message)} for channel {channel_name}")
@@ -442,6 +526,11 @@ class FjerkroaBot(commands.Bot):
 
     async def handle_message_through_responder(self, message):
         """Handle a message through the AI responder"""
+        # Ignored channels are fully silent — before the classifier gate,
+        # so no emoji reaction leaks either (BEH-09). DMs are never ignored.
+        if not isinstance(message.channel, DMChannel) and self.channel_ignored(self.get_channel_name(message.channel)):
+            self.log_message_action("ignore", message, self.get_channel_name(message.channel))
+            return
         message_content = str(message.content).strip()
         if message.reference and message.reference.resolved and isinstance(message.reference.resolved.content, str):
             reference_content = str(message.reference.resolved.content).replace("\n", "> \n")
@@ -612,6 +701,9 @@ class FjerkroaBot(commands.Bot):
         # Get the AI responder based on the channel name
         airesponder = self.get_ai_responder(channel_name)
 
+        # Classifier verdict rides along: factual questions may use factual-model (BEH-10)
+        message.factual = factual
+
         # Send the user message to the AI responder, with typing indicators.
         # A raised call = a broken API path (cf. the gpt-5.6 tools incident):
         # count it, alert staff at threshold, never crash the handler (OPS-16).

fjerkroa_bot/monitor.py

@@ -0,0 +1,82 @@
+"""Proactive health monitoring -> staff alerts (SPEC-012, FDB-012).
+
+A periodic check that watches daily spend against the budget, free disk,
+and task-queue depth, and posts a staff alert when a threshold is crossed
+— once per crossing, re-arming when the metric recovers, so a persistent
+condition never spams. Opt-in per deployment (`enable-monitoring`); it
+reuses the rate-limited staff-alert channel (OPS-07).
+"""
+
+import logging
+from typing import Any, Callable, Dict, Optional, Tuple
+
+# A check returns (metric-name, is-over-threshold, alert-message) or None when not applicable.
+Check = Optional[Tuple[str, bool, str]]
+
+
+class HealthMonitor:
+    def __init__(
+        self,
+        config_getter: Callable[[], Dict[str, Any]],
+        ledger: Any,
+        store: Any,
+        disk_free_mb: Callable[[], float],
+        alert: Callable[[str], Any],
+    ) -> None:
+        self._config = config_getter
+        self._ledger = ledger
+        self._store = store
+        self._disk_free_mb = disk_free_mb
+        self._alert = alert
+        self._armed: Dict[str, bool] = {}
+
+    def enabled(self) -> bool:
+        return bool(self._config().get("enable-monitoring", False))
+
+    def _check_spend(self) -> Check:
+        config = self._config()
+        if "daily-budget-usd" not in config:
+            return None
+        budget = float(config["daily-budget-usd"])
+        if budget <= 0:
+            return None
+        spent = float(self._ledger.spent_usd())
+        frac = spent / budget
+        threshold = float(config.get("monitor-spend-alert-frac", 0.8))
+        return ("spend", frac >= threshold, f"💸 Spend at ${spent:.2f} of ${budget:.2f} today ({frac:.0%}, alert ≥ {threshold:.0%}).")
+
+    def _check_disk(self) -> Check:
+        try:
+            free = float(self._disk_free_mb())
+        except Exception as err:
+            logging.debug(f"monitor: disk check failed: {err!r}")
+            return None
+        min_mb = float(self._config().get("monitor-disk-min-mb", 500))
+        return ("disk", free < min_mb, f"💾 Low disk: {free:.0f} MB free (alert < {min_mb:.0f} MB).")
+
+    def _check_queue(self) -> Check:
+        if self._store is None:
+            return None
+        try:
+            depth = len(self._store.tasks_open())
+        except Exception as err:
+            logging.debug(f"monitor: queue check failed: {err!r}")
+            return None
+        limit = int(self._config().get("monitor-taskqueue-max", 20))
+        return ("task-queue", depth >= limit, f"🗒️ Task queue deep: {depth} open (alert ≥ {limit}).")
+
+    async def tick(self) -> None:
+        """Evaluate every check; alert on a rising edge only (OPS-18/19)."""
+        for check in (self._check_spend(), self._check_disk(), self._check_queue()):
+            if check is None:
+                continue
+            metric, over, message = check
+            await self._fire(metric, over, message)
+
+    async def _fire(self, metric: str, over: bool, message: str) -> None:
+        was_over = self._armed.get(metric, False)
+        if over and not was_over:
+            self._armed[metric] = True
+            await self._alert(message)
+        elif not over and was_over:
+            self._armed[metric] = False  # recovered — re-arm silently for the next crossing

fjerkroa_bot/news.py

@@ -11,8 +11,10 @@ CLI: python -m fjerkroa_bot.news --config kroa.toml
 
 import argparse
 import logging
+import re
 import sys
 import time
+from html import unescape
 from typing import Any, Dict, List, Optional, Tuple
 
 import defusedxml.ElementTree as ElementTree  # hardened XML: feeds are untrusted (XXE/billion-laughs)
@@ -21,44 +23,68 @@ from .ai_responder import sanitize_external_text
 
 DEFAULT_PER_FEED = 3
 DEFAULT_MAX_ITEMS = 15
+DEFAULT_SUMMARY_CHARS = 200
+DEFAULT_NEWS_KEEP = 400
 FETCH_TIMEOUT_S = 15
 _ATOM = "{http://www.w3.org/2005/Atom}"
+_RSS1 = "{http://purl.org/rss/1.0/}"  # RSS 1.0 / RDF (e.g. 4gamer.net) namespaces <item>/<title>/<link>
+_TAG_RE = re.compile(r"<[^>]+>")
+
+
+def _clean_summary(raw: str, max_len: int = 300) -> str:
+    """Strip HTML, unescape entities, collapse whitespace (feed descriptions are often HTML)."""
+    text = unescape(_TAG_RE.sub(" ", raw or ""))
+    return re.sub(r"\s+", " ", text).strip()[:max_len]
+
+
+def _rss_items(root: Any, ns: str, source: str) -> List[Dict[str, str]]:
+    """RSS 2.0 (ns='') and RSS 1.0/RDF (ns=_RSS1) both use <item><title><link><description>."""
+    out: List[Dict[str, str]] = []
+    for item in root.iter(f"{ns}item"):
+        title = (item.findtext(f"{ns}title") or "").strip()
+        link = (item.findtext(f"{ns}link") or "").strip()
+        summary = _clean_summary(item.findtext(f"{ns}description") or "")
+        if title:
+            out.append({"title": title, "link": link, "source": source, "summary": summary})
+    return out
 
 
 def parse_feed(data: bytes, source: str = "") -> List[Dict[str, str]]:
-    """Parse RSS or Atom bytes into [{title, link, source}] (tolerant)."""
+    """Parse RSS 2.0, RSS 1.0/RDF, or Atom bytes into [{title, link, source, summary}] (tolerant)."""
     try:
         root = ElementTree.fromstring(data)
     except Exception as err:
         # malformed XML or a blocked entity/DTD attack — tolerate, never raise (NEWS-01)
         logging.warning(f"news: unparseable/unsafe feed {source!r}: {err!r}")
         return []
-    items: List[Dict[str, str]] = []
-    # RSS: <rss><channel><item><title/><link/>
-    for item in root.iter("item"):
-        title = (item.findtext("title") or "").strip()
-        link = (item.findtext("link") or "").strip()
-        if title:
-            items.append({"title": title, "link": link, "source": source})
-    # Atom: <feed><entry><title/><link href=/>
+    # RSS 2.0 (unqualified) + RSS 1.0/RDF (namespaced, e.g. 4gamer) share <item><title><link><description>
+    items: List[Dict[str, str]] = _rss_items(root, "", source) + _rss_items(root, _RSS1, source)
+    # Atom: <feed><entry><title/><link href=/><summary|content/>
     for entry in root.iter(f"{_ATOM}entry"):
         title = (entry.findtext(f"{_ATOM}title") or "").strip()
         link_el = entry.find(f"{_ATOM}link")
         link = link_el.get("href", "") if link_el is not None else ""
+        summary = _clean_summary(entry.findtext(f"{_ATOM}summary") or entry.findtext(f"{_ATOM}content") or "")
         if title:
-            items.append({"title": title, "link": link, "source": source})
+            items.append({"title": title, "link": link, "source": source, "summary": summary})
     return items
 
 
-def render_digest(items: List[Dict[str, str]], max_items: int = DEFAULT_MAX_ITEMS) -> str:
-    """Compact sanitized digest for the {news} prompt slot."""
+def render_digest(items: List[Dict[str, str]], max_items: int = DEFAULT_MAX_ITEMS, summary_chars: int = DEFAULT_SUMMARY_CHARS) -> str:
+    """Compact sanitized digest for the {news} prompt slot (title + short summary + link)."""
     lines = []
     for item in items[:max_items]:
         title = sanitize_external_text(item["title"], 200)
         source = item.get("source", "")
         link = item.get("link", "")
+        summary = sanitize_external_text(item.get("summary", ""), summary_chars) if summary_chars else ""
         prefix = f"[{source}] " if source else ""
-        lines.append(f"- {prefix}{title}" + (f" ({link})" if link else ""))
+        line = f"- {prefix}{title}"
+        if summary:
+            line += f" — {summary}"
+        if link:
+            line += f" ({link})"
+        lines.append(line)
     return "\n".join(lines)
 
 
@@ -102,10 +128,11 @@ def item_key(item: Dict[str, str]) -> str:
 class NewsPoster:
     """Post NEW feed items to Discord channel webhooks (ggg model, SPEC-013 NEWS-04..06)."""
 
-    def __init__(self, guard, fetch_bytes, post_webhook) -> None:
+    def __init__(self, guard, fetch_bytes, post_webhook, store: Any = None) -> None:
         self._guard = guard
         self._fetch_bytes = fetch_bytes
         self._post_webhook = post_webhook
+        self._store = store
 
     async def run_post(
         self,
@@ -115,9 +142,11 @@ class NewsPoster:
         per_feed: int,
         max_per_run: int,
         seed_only: bool,
+        keep: int = DEFAULT_NEWS_KEEP,
     ) -> Tuple[int, set]:
         """Returns (posted_count, updated_seen). seed_only marks new items seen without posting."""
         posted = 0
+        harvested: List[Dict[str, str]] = []
         for url, label, channel in feeds:
             reason = self._guard(url)
             if reason:
@@ -129,6 +158,7 @@ class NewsPoster:
                 logging.warning(f"news-post: fetch failed for {label}: {repr(err)}")
                 continue
             for item in parse_feed(data, label)[:per_feed]:
+                harvested.append(item)  # NEWS-09: everything parsed feeds the searchable store
                 key = item_key(item)
                 if not key or key in seen:
                     continue
@@ -136,6 +166,9 @@ class NewsPoster:
                 may_post = not seed_only and posted < max_per_run
                 if may_post and await self._deliver(item, label, channel, webhooks):
                     posted += 1
+        if self._store is not None and harvested:
+            self._store.add_news_items(harvested)
+            self._store.prune_news(keep)
         return posted, seen
 
     async def _deliver(self, item: Dict[str, str], label: str, channel: str, webhooks: Dict[str, str]) -> bool:
@@ -154,6 +187,92 @@ class NewsPoster:
             return False
 
 
+# --- news memory + on-demand retrieval tool (SPEC-013 NEWS-09..12) ---
+
+GET_NEWS_TOOL = {
+    "name": "get_news",
+    "description": "Fetch news the bot has collected from its RSS feeds — this is the SAME news that gets posted in the "
+    "server's news channels (e.g. #news, #newsjp / ニュース). Use this FIRST, before web_search, for anything about "
+    "current news or about something someone saw in a news channel; filter by topic (a keyword, also matches the source "
+    "label) or by source. Returns headlines with a short summary and a link; follow up with fetch_url for the full text.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "topic": {
+                "type": "string",
+                "description": "Optional filter: one or two keywords, in the language the feeds are written in "
+                "(e.g. Norwegian for Norwegian news: 'Nordland', 'fotball', 'trafikkulykke'). If nothing matches "
+                "exactly, related or recent items come back with a `note` saying so.",
+            },
+            "source": {"type": "string", "description": "Optional source label, e.g. 'NRK', 'Aftenposten', 'Verden', 'Sport'."},
+            "limit": {"type": "integer", "description": "How many items to return (default 10, max 30)."},
+        },
+        "required": [],
+    },
+}
+
+
+def _news_terms(topic: Optional[str]) -> List[str]:
+    return [t for t in re.split(r"\W+", (topic or "").lower()) if len(t) > 1][:5]
+
+
+def query_news(
+    store: Any, topic: Optional[str] = None, source: Optional[str] = None, limit: int = 10, summary_chars: int = DEFAULT_SUMMARY_CHARS
+) -> Dict[str, Any]:
+    """Retrieve stored news for the get_news tool: term/source-filtered, sanitized (NEWS-11)."""
+    if store is None:
+        return {"error": "news store unavailable"}
+    limit = max(1, min(int(limit or 10), 30))
+    src = (str(source).strip() or None) if source else None
+    terms = _news_terms(topic)
+    note = None
+    try:
+        rows = store.search_news(terms, limit, src) if terms else store.recent_news(limit, src)
+        if terms and not rows:  # NEWS-13: soft degradation, never empty-handed
+            rows = store.search_news(terms, limit, src, match_any=True)
+            note = "no item matches all keywords; showing items matching some of them"
+        if terms and not rows:
+            rows = store.recent_news(limit, src)
+            note = "nothing matches the topic; showing the newest stored items instead"
+    except Exception as err:
+        logging.warning(f"news: query failed: {err!r}")
+        return {"error": "news lookup failed"}
+    results = [
+        {
+            "source": row.get("source", ""),
+            "title": sanitize_external_text(row.get("title", ""), 200),
+            "summary": sanitize_external_text(row.get("summary", ""), summary_chars),
+            "link": row.get("link", ""),
+        }
+        for row in rows
+    ]
+    payload = {"topic": topic or "", "source": src or "", "results": results}
+    if note:
+        payload["note"] = note
+    return payload
+
+
+def _open_store(config: Dict[str, Any]) -> Any:
+    directory = config.get("history-directory")
+    if not directory:
+        return None
+    from pathlib import Path
+
+    from .persistence import PersistentStore
+
+    return PersistentStore(Path(str(directory)).expanduser() / "bot.db")
+
+
+def persist_news(config: Dict[str, Any], items: List[Dict[str, str]]) -> int:
+    """Upsert fetched items into the news store, prune to the rolling window (NEWS-09)."""
+    store = _open_store(config)
+    if store is None or not items:
+        return 0
+    added = store.add_news_items(items)
+    store.prune_news(int(config.get("news-keep", DEFAULT_NEWS_KEEP)))
+    return added
+
+
 def load_seen(path: str) -> Tuple[set, bool]:
     """(seen-set, existed). Missing/broken state -> empty set, existed=False (seed run)."""
     import json
@@ -231,7 +350,7 @@ async def run_post(config: Dict[str, Any]) -> int:
         logging.error("news-post: need news-post-webhooks and news-post-feeds")
         return 0
     seen, existed = load_seen(state_path)
-    poster = NewsPoster(guard_url, _aiohttp_fetch, _aiohttp_post)
+    poster = NewsPoster(guard_url, _aiohttp_fetch, _aiohttp_post, store=_open_store(config))
     posted, seen = await poster.run_post(
         feeds,
         webhooks,
@@ -239,6 +358,7 @@ async def run_post(config: Dict[str, Any]) -> int:
         int(config.get("news-post-per-feed", DEFAULT_POST_PER_FEED)),
         int(config.get("news-post-max-per-run", DEFAULT_POST_MAX_PER_RUN)),
         seed_only=not existed,  # first run seeds without flooding the channels
+        keep=int(config.get("news-keep", DEFAULT_NEWS_KEEP)),
     )
     save_seen(state_path, seen, int(config.get("news-post-seen-cap", DEFAULT_SEEN_CAP)))
     logging.info(f"news-post: posted {posted} item(s)" + (" (seed run — nothing posted)" if not existed else ""))
@@ -258,7 +378,10 @@ async def run(config: Dict[str, Any]) -> Optional[str]:
         return None
     fetcher = NewsFetcher(guard_url, _aiohttp_fetch)
     items = await fetcher.collect(feeds, int(config.get("news-per-feed", DEFAULT_PER_FEED)))
-    digest = render_digest(items, int(config.get("news-max-items", DEFAULT_MAX_ITEMS)))
+    persist_news(config, items)  # NEWS-09: feed the searchable rolling store for get_news
+    digest = render_digest(
+        items, int(config.get("news-max-items", DEFAULT_MAX_ITEMS)), int(config.get("news-summary-chars", DEFAULT_SUMMARY_CHARS))
+    )
     header = f"News as of {time.strftime('%Y-%m-%d %H:%M UTC', time.gmtime())}:\n"
     with open(out_path, "w", encoding="utf-8") as fd:
         fd.write(header + digest + "\n")

fjerkroa_bot/openai_responder.py

@@ -14,8 +14,12 @@ from .codex import DEFAULT_LIMIT as CODEX_DEFAULT_LIMIT
 from .codex import CodexSearch
 from .igdblib import IGDBQuery
 from .leonardo_draw import LeonardoAIDrawMixIn
+from .news import GET_NEWS_TOOL, query_news
 from .quota import QuotaLedger
 from .url_reader import FETCH_URL_TOOL, URLReader
+from .weather import GET_WEATHER_TOOL, Weather
+from .websearch import DEFAULT_RESULTS as WEB_DEFAULT_RESULTS
+from .websearch import WEB_SEARCH_TOOL, WebSearch
 
 # The response envelope, enforced server-side via structured outputs
 # (ENV-19). All fields required, closed object, nullable where the
@@ -36,6 +40,9 @@ ENVELOPE_SCHEMA = {
     "additionalProperties": False,
 }
 ENVELOPE_RESPONSE_FORMAT = {"type": "json_schema", "json_schema": {"name": "envelope", "strict": True, "schema": ENVELOPE_SCHEMA}}
+# Same schema in the Responses API shape (ENV-22): text.format is flat, not nested under json_schema
+ENVELOPE_TEXT_FORMAT = {"format": {"type": "json_schema", "name": "envelope", "strict": True, "schema": ENVELOPE_SCHEMA}}
+DEFAULT_RESPONSES_TOOL_ROUNDS = 4
 
 # Consolidation output (SPEC-002 MEM-02/03): new self-authored facts + one episode summary
 CONSOLIDATION_SCHEMA = {
@@ -121,6 +128,10 @@ async def openai_chat(client, *args, **kwargs):
     return await client.chat.completions.create(*args, **kwargs)
 
 
+async def openai_responses(client, *args, **kwargs):
+    return await client.responses.create(*args, **kwargs)
+
+
 async def openai_image(client, *args, **kwargs):
     return await client.images.generate(*args, **kwargs)
 
@@ -165,6 +176,9 @@ class OpenAIResponder(AIResponder, LeonardoAIDrawMixIn):
         self.url_reader = URLReader(lambda: self.config, self.image_cache)
         # Codex Mechanicus search (SPEC-014); Luma's own archive at binaric.tech
         self.codex = CodexSearch(lambda: self.config)
+        # Web search (SPEC-015) via Exa; general "look it up" beyond fetch_url/news/codex
+        self.web_search = WebSearch(lambda: self.config)
+        self.weather = Weather(lambda: self.config)
 
     def _available_tools(self) -> List[Dict[str, Any]]:
         """Assemble the function-tool list from every enabled provider (URL-01)."""
@@ -180,6 +194,12 @@ class OpenAIResponder(AIResponder, LeonardoAIDrawMixIn):
             functions.append(FETCH_URL_TOOL)
         if self.codex.enabled():  # CDX-01
             functions.append(CODEX_SEARCH_TOOL)
+        if self.config.get("enable-news-tool", False) and self.store is not None:  # NEWS-10
+            functions.append(GET_NEWS_TOOL)
+        if self.web_search.enabled():  # WEB-01
+            functions.append(WEB_SEARCH_TOOL)
+        if self.weather.enabled():  # WEA-01
+            functions.append(GET_WEATHER_TOOL)
         return functions
 
     async def _dispatch_tool(self, name: str, args: Dict[str, Any], author: str) -> Any:
@@ -197,6 +217,25 @@ class OpenAIResponder(AIResponder, LeonardoAIDrawMixIn):
             self.ledger._add(f"codex:{author}", 1)
             limit = int(self.config.get("codex-limit", CODEX_DEFAULT_LIMIT))
             return await self.codex.search(str(args.get("query", "")), str(args.get("lang", "en")), limit)
+        if name == "get_news":
+            per_user_cap = int(self.config.get("news-daily-per-user", 30))
+            if self.ledger._get(f"news:{author}") >= per_user_cap:  # NEWS-12
+                return {"error": "daily news lookup limit reached"}
+            self.ledger._add(f"news:{author}", 1)
+            summary_chars = int(self.config.get("news-summary-chars", 200))
+            return query_news(self.store, args.get("topic"), args.get("source"), args.get("limit", 10), summary_chars)
+        if name == "web_search":
+            per_user_cap = int(self.config.get("web-daily-per-user", 30))
+            if self.ledger._get(f"web:{author}") >= per_user_cap:  # WEB-05
+                return {"error": "daily web search limit reached"}
+            self.ledger._add(f"web:{author}", 1)
+            return await self.web_search.search(str(args.get("query", "")), int(args.get("num_results", WEB_DEFAULT_RESULTS)))
+        if name == "get_weather":
+            per_user_cap = int(self.config.get("weather-daily-per-user", 30))
+            if self.ledger._get(f"weather:{author}") >= per_user_cap:  # WEA-04
+                return {"error": "daily weather lookup limit reached"}
+            self.ledger._add(f"weather:{author}", 1)
+            return await self.weather.forecast(args.get("location"))
         return await self._execute_igdb_function(name, args)
 
     async def draw_openai(self, description: str, count: int = 1) -> List[BytesIO]:
@@ -237,9 +276,128 @@ class OpenAIResponder(AIResponder, LeonardoAIDrawMixIn):
         usage = getattr(result, "usage", None)
         prompt_tokens = getattr(usage, "prompt_tokens", None)
         completion_tokens = getattr(usage, "completion_tokens", None)
+        if not isinstance(prompt_tokens, int):  # Responses API names them input/output (ENV-22)
+            prompt_tokens = getattr(usage, "input_tokens", None)
+        if not isinstance(completion_tokens, int):
+            completion_tokens = getattr(usage, "output_tokens", None)
         if isinstance(prompt_tokens, int) and isinstance(completion_tokens, int):
             self.ledger.add_tokens(prompt_tokens, completion_tokens)
 
+    @staticmethod
+    def _responses_input(messages: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+        """Chat-format history -> Responses input items; vision parts become input_image (ENV-22)."""
+        items: List[Dict[str, Any]] = []
+        for msg in messages:
+            role = msg.get("role")
+            if role == "tool":
+                continue
+            content = msg.get("content")
+            if isinstance(content, list):
+                parts: List[Dict[str, Any]] = []
+                for part in content:
+                    if part.get("type") == "text":
+                        parts.append({"type": "input_text", "text": part.get("text", "")})
+                    elif part.get("type") == "image_url":
+                        parts.append({"type": "input_image", "image_url": part.get("image_url", {}).get("url", "")})
+                items.append({"role": role, "content": parts})
+            else:
+                items.append({"role": role, "content": str(content)})
+        return items
+
+    # Only these item types travel back as input; response-only fields like `status`
+    # are rejected by the API as unknown parameters (live 400, 2026-07-17)
+    _RESPONSES_FEEDBACK_FIELDS = {
+        "reasoning": ("id", "summary", "encrypted_content"),
+        "function_call": ("id", "call_id", "name", "arguments"),
+    }
+
+    @classmethod
+    def _responses_feedback(cls, output: List[Any]) -> List[Dict[str, Any]]:
+        """Reasoning + function_call items in input shape — keeps the chain of thought (ENV-23)."""
+        items: List[Dict[str, Any]] = []
+        for item in output or []:
+            fields = cls._RESPONSES_FEEDBACK_FIELDS.get(getattr(item, "type", None) or "")
+            if not fields:
+                continue  # message items need not travel back
+            data: Dict[str, Any] = {"type": item.type}
+            for field in fields:
+                value = getattr(item, field, None)
+                if field == "summary" and isinstance(value, list):
+                    value = [part if isinstance(part, dict) else part.model_dump() for part in value]
+                if value is not None:
+                    data[field] = value
+            items.append(data)
+        return items
+
+    @staticmethod
+    def _responses_refused(result: Any) -> bool:
+        for item in getattr(result, "output", []) or []:
+            if getattr(item, "type", None) == "message":
+                for part in getattr(item, "content", []) or []:
+                    if getattr(part, "type", None) == "refusal":
+                        return True
+        return False
+
+    async def _chat_via_responses(self, messages: List[Dict[str, Any]], limit: int, model: str) -> Tuple[Optional[Dict[str, Any]], int]:
+        """Responder call via /v1/responses: tools + reasoning allowed, stateless with encrypted reasoning (ENV-22/23)."""
+        context: List[Any] = self._responses_input(messages)
+        kwargs: Dict[str, Any] = {
+            "model": model,
+            "input": context,
+            "text": ENVELOPE_TEXT_FORMAT,
+            "store": False,  # nothing retained server-side (ENV-23)
+            "include": ["reasoning.encrypted_content"],
+            "reasoning": {"effort": str(self.config.get("reasoning-effort", "none"))},
+        }
+        author = self._last_author(messages)
+        if author:
+            # hashed, never the raw Discord name (SAF-10)
+            kwargs["safety_identifier"] = "discord-" + hashlib.sha256(author.encode()).hexdigest()[:16]
+        available_tools = self._available_tools()
+        if available_tools:
+            kwargs["tools"] = [{"type": "function", **func} for func in available_tools]
+            kwargs["tool_choice"] = "auto"
+            logging.info(f"🔧 Tools available to AI: {[func['name'] for func in available_tools]}")
+
+        rounds = int(self.config.get("responses-tool-rounds", DEFAULT_RESPONSES_TOOL_ROUNDS))
+        for _ in range(max(1, rounds) + 1):
+            result = await openai_responses(self.client, **kwargs)
+            self._record_usage(result)
+            if self._responses_refused(result):
+                logging.warning("model refused (responses path)")  # ENV-24
+                return None, limit
+            calls = [item for item in (getattr(result, "output", []) or []) if getattr(item, "type", None) == "function_call"]
+            if not calls or "tools" not in kwargs:
+                answer = {"content": getattr(result, "output_text", None) or "", "role": "assistant"}
+                self.rate_limit_backoff = exponential_backoff()
+                self._use_retry_model = False
+                logging.info(f"generated response {getattr(result, 'usage', None)}: {repr(answer)}")
+                return answer, limit
+            tool_names = [call.name for call in calls]
+            logging.info(f"🔧 OpenAI requested function calls: {tool_names}")
+            # Pass reasoning + function_call items back — keeps the chain of thought (ENV-23)
+            context = context + self._responses_feedback(result.output)
+            for call in calls:
+                function_args = json.loads(call.arguments) if call.arguments else {}
+                logging.info(f"🔧 Executing tool: {call.name} with args: {function_args}")
+                function_result = await self._dispatch_tool(call.name, function_args, author or "")
+                logging.info(f"🔧 Tool result: {type(function_result)} - {str(function_result)[:200]}...")
+                context.append(
+                    {
+                        "type": "function_call_output",
+                        "call_id": call.call_id,
+                        # tool text is external input — sanitize before prompting (SAF-03)
+                        "output": sanitize_external_text(json.dumps(function_result), 8000) if function_result else "No results found",
+                    }
+                )
+            kwargs["input"] = context
+            rounds -= 1
+            if rounds <= 0:
+                # loop exhausted: force a tool-less final answer (ENV-23)
+                kwargs.pop("tools", None)
+                kwargs.pop("tool_choice", None)
+        return None, limit
+
     async def chat(self, messages: List[Dict[str, Any]], limit: int) -> Tuple[Optional[Dict[str, Any]], int]:
         # Safety check for mock objects in tests
         if not isinstance(messages, list) or len(messages) == 0:
@@ -270,12 +428,17 @@ class OpenAIResponder(AIResponder, LeonardoAIDrawMixIn):
                 model = self.config["model-vision"]
             else:
                 messages[-1]["content"] = messages[-1]["content"][0]["text"]
+            if getattr(self, "_factual", False) and "factual-model" in self.config:
+                model = self.config["factual-model"]  # BEH-10: facts get the stronger tier
             if self._use_retry_model and "retry-model" in self.config:
                 model = self.config["retry-model"]
         except (KeyError, IndexError, TypeError) as e:
             logging.warning(f"Error accessing message content: {e}")
             return None, limit
         try:
+            if bool(self.config.get("use-responses-api", False)):
+                return await self._chat_via_responses(messages, limit, model)  # ENV-22
+
             # Prepare function calls if IGDB is enabled
             chat_kwargs = {
                 "model": model,

fjerkroa_bot/persistence.py

@@ -13,7 +13,7 @@ from contextlib import closing
 from pathlib import Path
 from typing import Any, Dict, List, Optional
 
-SCHEMA_VERSION = 5
+SCHEMA_VERSION = 6
 
 
 class PersistentStore:
@@ -72,6 +72,13 @@ class PersistentStore:
                     " due_at TEXT NOT NULL, payload TEXT NOT NULL, state TEXT NOT NULL DEFAULT 'queued',"
                     " created_at TEXT NOT NULL DEFAULT (datetime('now')), executed_at TEXT)"
                 )
+            if version < 6:
+                # News memory (SPEC-013 NEWS-09): deduped rolling store of fetched items
+                conn.execute(
+                    "CREATE TABLE IF NOT EXISTS news (id INTEGER PRIMARY KEY, dedup_key TEXT UNIQUE NOT NULL,"
+                    " source TEXT NOT NULL DEFAULT '', title TEXT NOT NULL, link TEXT NOT NULL DEFAULT '',"
+                    " summary TEXT NOT NULL DEFAULT '', first_seen TEXT NOT NULL DEFAULT (datetime('now')))"
+                )
             if version < SCHEMA_VERSION:
                 conn.execute(f"PRAGMA user_version = {SCHEMA_VERSION}")
         os.chmod(self.db_path, 0o600)  # conversation data (PER-04)
@@ -115,6 +122,77 @@ class PersistentStore:
             row = conn.execute("SELECT value FROM usage WHERE day = ? AND key = ?", (day, key)).fetchone()
         return float(row[0]) if row else 0.0
 
+    # --- news memory (SPEC-013 NEWS-09..12) ---
+
+    def add_news_items(self, items: List[Dict[str, Any]]) -> int:
+        """Insert deduped news rows (by link or title); returns how many were new (NEWS-09)."""
+        added = 0
+        with closing(self._connect()) as conn, conn:
+            for item in items:
+                title = str(item.get("title") or "").strip()
+                key = (str(item.get("link") or "").strip()) or title
+                if not title or not key:
+                    continue
+                cursor = conn.execute(
+                    "INSERT OR IGNORE INTO news (dedup_key, source, title, link, summary) VALUES (?, ?, ?, ?, ?)",
+                    (key, str(item.get("source") or ""), title, str(item.get("link") or ""), str(item.get("summary") or "")),
+                )
+                added += cursor.rowcount
+        return added
+
+    def recent_news(self, limit: int = 20, source: Optional[str] = None) -> List[Dict[str, Any]]:
+        sql = "SELECT source, title, link, summary FROM news"
+        params: List[Any] = []
+        if source:
+            sql += " WHERE source = ?"
+            params.append(source)
+        sql += " ORDER BY id DESC LIMIT ?"
+        params.append(int(limit))
+        with closing(self._connect()) as conn:
+            rows = conn.execute(sql, params).fetchall()
+        return [{"source": r[0], "title": r[1], "link": r[2], "summary": r[3]} for r in rows]
+
+    def search_news(self, terms: List[str], limit: int = 20, source: Optional[str] = None, match_any: bool = False) -> List[Dict[str, Any]]:
+        """Rows where every term appears in title/summary/source; match_any ranks by how many terms hit (NEWS-11/13)."""
+        params: List[Any] = []
+        clauses = []
+        for term in terms:
+            clauses.append("(title LIKE ? OR summary LIKE ? OR source LIKE ?)")
+            like = f"%{term}%"
+            params += [like, like, like]
+        if match_any and clauses:
+            hits = " + ".join(clauses)
+            where = "hits > 0"
+            if source:
+                where += " AND source = ?"
+                params.append(source)
+            params.append(int(limit))
+            sql = (
+                f"SELECT source, title, link, summary FROM "  # nosec B608 - fixed templates; values parameterised
+                f"(SELECT id, source, title, link, summary, {hits} AS hits FROM news) "
+                f"WHERE {where} ORDER BY hits DESC, id DESC LIMIT ?"
+            )
+        else:
+            where = " AND ".join(clauses) if clauses else "1=1"
+            if source:
+                where = f"({where}) AND source = ?"
+                params.append(source)
+            params.append(int(limit))
+            sql = f"SELECT source, title, link, summary FROM news WHERE {where} ORDER BY id DESC LIMIT ?"  # nosec B608 - fixed templates; values parameterised
+        with closing(self._connect()) as conn:
+            rows = conn.execute(sql, params).fetchall()
+        return [{"source": r[0], "title": r[1], "link": r[2], "summary": r[3]} for r in rows]
+
+    def prune_news(self, keep: int) -> int:
+        """Keep the newest `keep` rows, delete the rest (rolling window, NEWS-09)."""
+        with closing(self._connect()) as conn, conn:
+            cursor = conn.execute("DELETE FROM news WHERE id NOT IN (SELECT id FROM news ORDER BY id DESC LIMIT ?)", (int(keep),))
+        return cursor.rowcount
+
+    def news_count(self) -> int:
+        with closing(self._connect()) as conn:
+            return int(conn.execute("SELECT COUNT(*) FROM news").fetchone()[0])
+
     def delete_history_of_user(self, user: str) -> int:
         """Remove persisted rows carrying this user's messages (SAF-08)."""
         with closing(self._connect()) as conn, conn:

fjerkroa_bot/tasks.py

@@ -17,7 +17,13 @@ from .quota import QuotaLedger
 DEFAULT_MAX_PER_CHANNEL_PER_DAY = 2
 DEFAULT_IDLE_IMPULSE_HOURS = 12.0
 DEFAULT_TASKGEN_INTERVAL_HOURS = 6.0
-DEFAULT_BORENESS_PROMPT = "Pretend that you just now thought of something, be creative."
+DEFAULT_BORENESS_PROMPT = (
+    "A thought just occurred to you. Anchor it to something real you know — recent news (use get_news), a game "
+    "releasing soon, the weather, or a regular you remember — not a generic musing. Share it briefly, in your own "
+    "voice, as an observation, a gentle question, or a joke; never an advertisement. Read the room and stay in character. "
+    "Check your own recent posts in the history first: pick a subject you have not touched lately and a different form "
+    "than last time, and never open with a fixed label or heading — just start mid-thought."
+)
 
 ExecuteCallback = Callable[[str, str], Awaitable[None]]
 ProposeCallback = Callable[[], Awaitable[Optional[Dict[str, Any]]]]

fjerkroa_bot/url_reader.py

@@ -21,7 +21,7 @@ from .ai_responder import sanitize_external_text
 from .httpread import read_capped
 
 DEFAULT_MAX_BYTES = 2 * 1024 * 1024
-DEFAULT_MAX_CHARS = 6000
+DEFAULT_MAX_CHARS = 8000  # URL-08: budget goes to content now, not chrome
 DEFAULT_MAX_IMAGES = 2
 FETCH_TIMEOUT_S = 15
 MAX_REDIRECTS = 5
@@ -41,18 +41,37 @@ FETCH_URL_TOOL = {
 _META_REFRESH_URL = re.compile(r"url\s*=\s*['\"]?([^'\";\s]+)", re.I)
 
 
+_SKIP_TAGS = ("script", "style", "noscript", "svg", "nav", "header", "footer", "aside", "form", "select", "button")
+_BLOCK_TAGS = ("p", "li", "div", "section", "article", "td", "ul", "ol", "table", "h1", "h2", "h3", "h4", "h5", "h6")
+_LINK_DENSITY_MAX = 0.6  # boilerplate: block mostly link text ... (URL-08)
+_LINK_BLOCK_MAX_CHARS = 200  # ... AND short (menus, related lists); long linky paragraphs survive
+
+
 class _Extractor(HTMLParser):
     def __init__(self) -> None:
         super().__init__()
         self._skip = 0
-        self.parts: List[str] = []
+        self._links = 0
+        self._buf: List[str] = []
+        self._buf_link_chars = 0
+        self.blocks: List[Tuple[str, int]] = []  # (text, chars inside <a>)
         self.images: List[str] = []
         self.og_image: Optional[str] = None
         self.refresh_url: Optional[str] = None
 
+    def _flush(self) -> None:
+        text = " ".join(self._buf).strip()
+        if text:
+            self.blocks.append((text, self._buf_link_chars))
+        self._buf, self._buf_link_chars = [], 0
+
     def handle_starttag(self, tag: str, attrs) -> None:
-        if tag in ("script", "style", "noscript", "svg"):
+        if tag in _SKIP_TAGS:
             self._skip += 1
+        if tag == "a":
+            self._links += 1
+        if tag in _BLOCK_TAGS:
+            self._flush()
         attr = dict(attrs)
         src = attr.get("src")
         if tag == "img" and src:
@@ -67,12 +86,28 @@ class _Extractor(HTMLParser):
                 self.refresh_url = match.group(1)
 
     def handle_endtag(self, tag: str) -> None:
-        if tag in ("script", "style", "noscript", "svg") and self._skip > 0:
+        if tag in _SKIP_TAGS and self._skip > 0:
             self._skip -= 1
+        if tag == "a" and self._links > 0:
+            self._links -= 1
+        if tag in _BLOCK_TAGS:
+            self._flush()
 
     def handle_data(self, data: str) -> None:
         if self._skip == 0 and data.strip():
-            self.parts.append(data.strip())
+            self._buf.append(data.strip())
+            if self._links > 0:
+                self._buf_link_chars += len(data.strip())
+
+    def content_parts(self) -> List[str]:
+        """Blocks minus boilerplate: short blocks dominated by link text are chrome (URL-08)."""
+        self._flush()
+        out = []
+        for text, link_chars in self.blocks:
+            if link_chars / max(1, len(text)) > _LINK_DENSITY_MAX and len(text) < _LINK_BLOCK_MAX_CHARS:
+                continue
+            out.append(text)
+        return out
 
 
 def _ip_is_public(ip_str: str) -> bool:
@@ -162,7 +197,7 @@ class URLReader:
         return extractor
 
     def _to_text(self, html: str) -> str:
-        return re.sub(r"\s+\n", "\n", " ".join(self._extract(html).parts))
+        return re.sub(r"\s+\n", "\n", " ".join(self._extract(html).content_parts()))
 
     async def _ingest_images(self, html: str, base_url: str, channel: str, user: str) -> int:
         if self.image_cache is None:

fjerkroa_bot/weather.py

@@ -0,0 +1,111 @@
+"""Weather tool via MET Norway Locationforecast (SPEC-016).
+
+A `get_weather` function tool: both personas talk about weather (the
+sea over the skerries, rain on patch day) but had to guess it. The
+free api.met.no compact forecast grounds it. Locations are
+host-configured `[name, lat, lon]` entries — the model picks by name
+and never supplies coordinates or URLs, so there is no SSRF surface.
+"""
+
+import logging
+from typing import Any, Callable, Dict, List, Optional, Tuple
+
+import aiohttp
+
+from .ai_responder import sanitize_external_text
+
+MET_COMPACT_URL = "https://api.met.no/weatherapi/locationforecast/2.0/compact"
+USER_AGENT = "fjerkroa-discord-bot/3 (https://fjerkroa.no)"
+FETCH_TIMEOUT_S = 15
+FORECAST_POINT_INDICES = (6, 12, 24)  # hourly series: ~6h/12h/24h ahead
+
+GET_WEATHER_TOOL = {
+    "name": "get_weather",
+    "description": "Current weather and a short forecast for the configured local places. Use this whenever weather comes "
+    "up in conversation — never guess or invent weather. Returns current temperature (°C), wind (m/s) and conditions, "
+    "plus a few forecast points.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "location": {"type": "string", "description": "Place name to look up; omit for the default (first configured) place."},
+        },
+        "required": [],
+    },
+}
+
+
+def _reduce(data: Any, name: str) -> Dict[str, Any]:
+    """Compact MET timeseries -> {location, now, forecast[]} (WEA-02). Nothing else reaches the prompt."""
+    series = data.get("properties", {}).get("timeseries", []) if isinstance(data, dict) else []
+    if not series:
+        return {"error": "weather data unavailable"}
+
+    def point(entry: Dict[str, Any]) -> Dict[str, Any]:
+        details = entry.get("data", {}).get("instant", {}).get("details", {})
+        hour = entry.get("data", {}).get("next_1_hours", {}) or entry.get("data", {}).get("next_6_hours", {})
+        out: Dict[str, Any] = {
+            "time": str(entry.get("time", "")),
+            "temp_c": details.get("air_temperature"),
+            "wind_ms": details.get("wind_speed"),
+        }
+        symbol = hour.get("summary", {}).get("symbol_code")
+        if symbol:
+            out["conditions"] = str(symbol)
+        precip = hour.get("details", {}).get("precipitation_amount")
+        if precip is not None:
+            out["precip_mm"] = precip
+        return out
+
+    forecast = [point(series[i]) for i in FORECAST_POINT_INDICES if i < len(series)]
+    return {"location": sanitize_external_text(name, 80), "now": point(series[0]), "forecast": forecast}
+
+
+class Weather:
+    def __init__(self, config_getter: Callable[[], Dict[str, Any]]) -> None:
+        self._config = config_getter
+
+    def _locations(self) -> List[Tuple[str, float, float]]:
+        out: List[Tuple[str, float, float]] = []
+        for entry in self._config().get("weather-locations", []):
+            try:
+                name, lat, lon = entry[0], float(entry[1]), float(entry[2])
+                out.append((str(name), lat, lon))
+            except (TypeError, ValueError, IndexError):
+                logging.warning(f"weather: bad location entry {entry!r}")
+        return out
+
+    def enabled(self) -> bool:
+        return bool(self._config().get("enable-weather", False)) and bool(self._locations())
+
+    def _pick(self, location: Optional[str]) -> Optional[Tuple[str, float, float]]:
+        """Case-insensitive substring match; unknown/absent = first configured (WEA-03)."""
+        entries = self._locations()
+        if not entries:
+            return None
+        wanted = (location or "").strip().casefold()
+        if wanted:
+            for entry in entries:
+                if wanted in entry[0].casefold():
+                    return entry
+        return entries[0]
+
+    async def _fetch_json(self, lat: float, lon: float) -> Any:
+        timeout = aiohttp.ClientTimeout(total=FETCH_TIMEOUT_S)
+        params = {"lat": f"{lat:.4f}", "lon": f"{lon:.4f}"}
+        async with aiohttp.ClientSession(timeout=timeout, headers={"User-Agent": USER_AGENT}) as session:
+            async with session.get(MET_COMPACT_URL, params=params) as response:
+                response.raise_for_status()
+                return await response.json()
+
+    async def forecast(self, location: Optional[str] = None) -> Dict[str, Any]:
+        """Return a compact forecast, or an error dict — never raise (WEA-04)."""
+        picked = self._pick(location)
+        if picked is None:
+            return {"error": "weather unavailable: no locations configured"}
+        name, lat, lon = picked
+        try:
+            data = await self._fetch_json(lat, lon)
+        except Exception as err:
+            logging.warning(f"weather fetch failed: {err!r}")
+            return {"error": "weather lookup failed"}
+        return _reduce(data, name)

fjerkroa_bot/websearch.py

@@ -0,0 +1,94 @@
+"""Web search tool via Exa (SPEC-015, FDB-022).
+
+A `web_search` function tool: the model looks things up on the open web
+when a general "look it up" question is not covered by IGDB, the codex,
+the news store, or a URL the user pasted. Results are external text, so
+titles and snippets are sanitized (SAF-03) before they reach the prompt.
+The Exa API key lives in host config (or the `EXA_API_KEY` env), never in
+the repo.
+"""
+
+import logging
+import os
+from typing import Any, Callable, Dict, List
+
+import aiohttp
+
+from .ai_responder import sanitize_external_text
+
+EXA_SEARCH_URL = "https://api.exa.ai/search"
+DEFAULT_RESULTS = 5
+MAX_RESULTS = 10
+DEFAULT_SNIPPET_CHARS = 400
+FETCH_TIMEOUT_S = 15
+
+WEB_SEARCH_TOOL = {
+    "name": "web_search",
+    "description": "Search the open web for general information. Use ONLY when the answer is not in your own sources: for "
+    "the server's news use get_news, for Adeptus Mechanicus / Warhammer 40k lore use codex_search, for video-game facts use "
+    "the game tools, for a specific URL someone pasted use fetch_url. Returns result titles, URLs, and a short snippet; "
+    "follow up with fetch_url on a result link for the full article.",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "query": {"type": "string", "description": "What to search the web for."},
+            "num_results": {"type": "integer", "description": "How many results to return (default 5, max 10)."},
+        },
+        "required": ["query"],
+    },
+}
+
+
+def _format_results(data: Any, snippet_chars: int) -> List[Dict[str, str]]:
+    """Reduce an Exa response to sanitized {title, url, snippet, published} rows (WEB-02)."""
+    results = data.get("results", []) if isinstance(data, dict) else []
+    out: List[Dict[str, str]] = []
+    for item in results:
+        if not isinstance(item, dict):
+            continue
+        out.append(
+            {
+                "title": sanitize_external_text(str(item.get("title") or ""), 200),
+                "url": str(item.get("url") or ""),
+                "snippet": sanitize_external_text(str(item.get("text") or item.get("snippet") or ""), snippet_chars),
+                "published": str(item.get("publishedDate") or ""),
+            }
+        )
+    return out
+
+
+class WebSearch:
+    def __init__(self, config_getter: Callable[[], Dict[str, Any]]) -> None:
+        self._config = config_getter
+
+    def _api_key(self) -> str:
+        return str(self._config().get("exa-api-key") or os.environ.get("EXA_API_KEY", ""))
+
+    def enabled(self) -> bool:
+        return bool(self._config().get("enable-web-search", False)) and bool(self._api_key())
+
+    async def _post(self, payload: Dict[str, Any], headers: Dict[str, str]) -> Any:
+        timeout = aiohttp.ClientTimeout(total=FETCH_TIMEOUT_S)
+        async with aiohttp.ClientSession(timeout=timeout) as session:
+            async with session.post(EXA_SEARCH_URL, json=payload, headers=headers) as response:
+                response.raise_for_status()
+                return await response.json()
+
+    async def search(self, query: str, num_results: int = DEFAULT_RESULTS) -> Dict[str, Any]:
+        """Return sanitized web results, or an error dict — never raise (WEB-04)."""
+        key = self._api_key()
+        if not key:
+            return {"error": "web search unavailable: no api key"}
+        query = (query or "").strip()
+        if not query:
+            return {"query": "", "results": []}
+        num = max(1, min(int(num_results or DEFAULT_RESULTS), MAX_RESULTS))  # WEB-03
+        snippet_chars = int(self._config().get("web-snippet-chars", DEFAULT_SNIPPET_CHARS))
+        payload = {"query": query, "numResults": num, "type": "auto", "contents": {"text": {"maxCharacters": max(snippet_chars, 200)}}}
+        headers = {"x-api-key": key, "Content-Type": "application/json"}
+        try:
+            data = await self._post(payload, headers)
+        except Exception as err:
+            logging.warning(f"web search failed: {err!r}")
+            return {"error": "web search failed"}
+        return {"query": query, "results": _format_results(data, snippet_chars)}

specs/SPEC-001-responder-envelope.md

@@ -152,3 +152,34 @@ Every chat call carries `response_format` = strict JSON schema named
 IMG-02), `picture_edit`, `hack` — all required,
 `additionalProperties: false`, nullable where the protocol allows
 null. Tool-followup calls carry the same format.
+
+### ENV-22 — Responses API path behind a flag (coverage: test)
+
+With `use-responses-api = true`, responder chat calls go to
+`/v1/responses` instead of chat/completions: same model selection
+(default / vision / factual / retry), the same strict envelope schema
+(as `text.format`), tools in the flat Responses shape, and
+`reasoning` = config `reasoning-effort` — tools + reasoning are
+allowed here (the chat/completions 400 from ENV-21 does not apply).
+Flag off (default) = the ENV-21 path, byte-identical behavior.
+Classifier, consolidation and task-proposal calls stay on
+chat/completions.
+
+### ENV-23 — Responses tool loop is stateless and keeps reasoning (coverage: test)
+
+The Responses path runs with `store=false` and
+`include=["reasoning.encrypted_content"]` (nothing retained
+server-side). On a function call, the reasoning and function_call
+output items are passed back as input — reduced to their input-shape
+fields, since response-only fields like `status` are rejected as
+unknown parameters (live 400, 2026-07-17) — together with one
+`function_call_output` per call (matched by `call_id`, result
+sanitized per SAF-03), so the model continues one chain of thought
+across tool rounds. Up to `responses-tool-rounds` (default 4) rounds
+may call tools; an exhausted loop forces a final tool-less answer.
+
+### ENV-24 — Responses refusals are failed attempts (coverage: test)
+
+A refusal content part in the Responses output yields no answer
+(backoff + retry per ENV-12/ENV-18), exactly like the
+chat/completions path.

specs/SPEC-006-operator-controls.md

@@ -73,3 +73,15 @@ per-channel) — without it, `!bot unpin <id>` required guessing ids.
 `!bot spend` answers in the staff channel with today's estimated
 spend in USD, token and image counts, and the configured budget.
 Management sees the cost, not just the cap.
+
+### OPS-17 — Help is complete and context-aware (coverage: test)
+
+Help reflects where each command actually works, because not every
+command is allowed everywhere. `!help` answers in any channel and
+lists only the commands usable there: in a normal channel the
+everyone-commands (`!help`, `!forgetme`, `!privacy`, `!wichtel`); in
+the staff channel it additionally lists the operator commands grouped
+by purpose (control, cost, memory, tasks). `!bot help` — and any
+unrecognised `!bot` command — answers with that same full staff help,
+so the listing is exhaustive rather than the old hand-maintained
+partial line. Help works even while the bot is paused.

specs/SPEC-008-config.md

@@ -31,3 +31,16 @@ and all responder `.config` references is scheduled onto the event
 loop (`call_soon_threadsafe`), so no request ever reads a
 half-swapped config (D9). Before the loop runs (startup), the swap
 applies directly — there are no concurrent readers yet.
+
+### CFG-05 — Hot-reload is rename-safe (coverage: test)
+
+The watcher observes the config file's **directory**, not the file, and
+reacts to a **modified, created, or moved** event whose source or
+destination path is the config file. This catches atomic saves — write
+a temp file, then rename it over the target — which replace the inode
+and fire a move/create rather than a modify; watching the file directly
+would go deaf after the first such save. Open/close events are
+deliberately not handled: reloading re-opens the file to read it, so
+reacting to opens would feed back into an endless reload loop. Events
+for other files in the directory, and directory events themselves, are
+ignored.

specs/SPEC-010-behavior.md

@@ -61,3 +61,24 @@ Within `quiet-hours = "HH:MM-HH:MM"` (host-local, may wrap midnight)
 `bot_initiated_allowed()` is false: no boreness, later no scheduler
 posts. Replies to users stay unaffected — a guest asking at 23:30
 still gets an answer.
+
+### BEH-09 — Ignored channels are fully silent (coverage: test)
+
+Channels matching `ignore-channels` get neither replies nor
+classifier emoji reactions: the message handler returns before the
+classifier gate, so no model call, no reaction, no history entry.
+Entries are fnmatch patterns (`todo*` matches `todo`, `todo-lists`);
+plain names keep matching exactly as before. DMs are never ignored.
+`channel_by_name` resolution honors the same patterns. (Previously
+the ignore check sat only in `respond()`, after the classifier —
+emoji reactions leaked into ignored channels, and matching was
+exact-name only.)
+
+### BEH-10 — Factual questions may use a stronger model (coverage: test)
+
+With `factual-model` configured, a message the classifier tagged
+`factual` (BEH-05) is answered by that model instead of `model` —
+opening hours, release dates, news lookups get the stronger tier
+while small talk stays on the cheap default. Unset = no change. The
+`retry-model` override still wins on retry, and vision inputs keep
+using `model-vision`.

specs/SPEC-011-url-reading.md

@@ -58,3 +58,14 @@ Each fetch increments a per-user daily counter; over
 `url-daily-per-user` (default 20) `fetch_url` refuses with an error
 result. The budget gate (SAF-04) still applies to the surrounding
 model calls.
+
+### URL-08 — Main-content extraction (coverage: test)
+
+`fetch_url` text drops page chrome: content inside
+`nav`/`header`/`footer`/`aside`/`form`/`select`/`button` is skipped
+like scripts, and text blocks dominated by link text (over 60 % of a
+block's characters inside `<a>` and the block shorter than 200 chars
+— menus, related-article lists, tag clouds) are treated as
+boilerplate and removed. Body paragraphs with inline links survive.
+The default `url-max-chars` cap rises to 8000 now that the budget is
+spent on content, not chrome.

specs/SPEC-012-ops.md

@@ -30,3 +30,23 @@ The responder counts consecutive OpenAI request failures; at
 alert (rate-limited like all staff alerts) so a silently-broken bot
 (cf. the gpt-5.6 tools/reasoning incident) surfaces within minutes
 instead of hours. A success resets the counter.
+
+### OPS-18 — Health monitor watches spend, disk, task-queue (coverage: test)
+
+When `enable-monitoring` is true, a loop wakes every `monitor-interval`
+(default 300 s) and checks three thresholds, alerting the staff channel
+when one is crossed: daily spend at or above `monitor-spend-alert-frac`
+(default 0.8) of `daily-budget-usd`; free disk below `monitor-disk-min-mb`
+(default 500 MB); open task-queue depth at or above `monitor-taskqueue-max`
+(default 20). A check with no data to evaluate (no budget set, no store,
+a failed disk read) is skipped, never fatal. With the flag off the loop
+does nothing.
+
+### OPS-19 — Alerts fire once per crossing and re-arm on recovery (coverage: test)
+
+Each metric alerts only on the rising edge — the first tick that finds
+it over its threshold — and stays silent while it remains over, so a
+persistent condition does not repeat every interval. When the metric
+falls back below the threshold the alert re-arms silently, ready to fire
+again on the next crossing. All alerts still pass through the
+rate-limited staff-alert path (OPS-07).

specs/SPEC-013-news.md

@@ -6,7 +6,7 @@ Replaces the broken pre-1.0-openai `news_feed.py`. A CLI
 `AIResponder.message` injects into the `{news}` slot. Feeds are
 external input and operator-configured.
 
-### NEWS-01 — RSS and Atom parse to items (coverage: test)
+### NEWS-01 — RSS (2.0 and 1.0/RDF) and Atom parse to items (coverage: test)
 
 `parse_feed(bytes, label)` extracts `{title, link, source}` from both
 RSS (`<item>`) and Atom (`<entry>`) documents, tolerates malformed
@@ -52,3 +52,60 @@ whose channel has no configured webhook, and a webhook POST that
 raises are each logged and skipped — one failure never sinks the
 run, and the seen-set still advances for successfully-processed
 items.
+
+### NEWS-07 — Item summaries are extracted (coverage: test)
+
+`parse_feed` also captures each item's short description — RSS
+`<description>`, Atom `<summary>` or `<content>` — with HTML stripped,
+entities unescaped, and whitespace collapsed, so an item carries what
+it is about, not only a headline. Missing descriptions yield an empty
+summary, never an error.
+
+### NEWS-08 — The digest carries summaries (coverage: test)
+
+`render_digest` appends the sanitized, length-capped
+(`news-summary-chars`, default 200) summary after each headline, so
+the bot's ambient `{news}` context knows the gist of each story, not
+just its title. A zero cap restores the title-only digest.
+
+### NEWS-09 — Fetched news is stored, deduped, and rolled over (coverage: test)
+
+Both the digest run (kroa) and the posting run (ggg) upsert every
+fetched item into a `news` table keyed by link (or title), so the same
+story is stored once. After each run the store is pruned to the newest
+`news-keep` rows (default 400), a rolling window that bounds growth
+while keeping recent history searchable.
+
+### NEWS-10 — get_news is offered as a tool (coverage: test)
+
+When `enable-news-tool` is true and a store is configured, the chat
+call's `tools` list includes a `get_news` function (optional `topic`,
+`source`, `limit`) next to the other tools. Without a store or the
+flag it is absent.
+
+### NEWS-11 — get_news retrieves filtered, sanitized items (coverage: test)
+
+`get_news` returns recent stored items, newest first, optionally
+narrowed by `topic` (every keyword must appear in the title, summary,
+or source label — so `topic: "Nordland"` finds items from that source)
+and/or an exact `source`; `limit` is clamped to 1..30. Each result's title and
+summary are passed through `sanitize_external_text`. The bot can then
+`fetch_url` a returned link for the full article.
+
+### NEWS-12 — get_news is metered per user (coverage: test)
+
+Each `get_news` call increments a per-user daily counter; over
+`news-daily-per-user` (default 30) the tool refuses with an error
+result without touching the store. The budget gate (SAF-04) still
+applies to the surrounding model calls.
+
+### NEWS-13 — Topic misses degrade softly, never empty-handed (coverage: test)
+
+A `topic` whose AND-match (NEWS-11) finds nothing falls back to an
+any-term match, ranked by how many keywords hit (ties: newest first);
+if that too is empty, the newest stored items are returned instead.
+Both fallbacks set a `note` field naming the degradation so the model
+can answer honestly ("nothing on that exactly, but…"). A model
+passing a multi-word or wrong-language topic (the live
+`"Nordland road accident"` → `[]` case) thus still gets usable
+context. Exact matches return no `note`.

specs/SPEC-014-codex.md

@@ -1,8 +1,8 @@
 # SPEC-014 — Codex Mechanicus search
 
-Luma is an Adeptus Mechanicus tech-priest; her lore has a real home —
+Luma is an Adeptus Mechanicus tech-priest; his lore has a real home —
 the priest's own Codex Mechanicus at `binaric.tech` (an Astro/MDX
-archive, five tongues). A `codex_search` function tool lets her consult
+archive, five tongues). A `codex_search` function tool lets him consult
 that archive and answer from sourced inscriptions instead of inventing
 lore. The index is public but still untrusted by the time it reaches a
 prompt: the fetch is SSRF-guarded (SPEC-011 shares `guard_url`),

specs/SPEC-015-websearch.md

@@ -0,0 +1,42 @@
+# SPEC-015 — Web search (Exa)
+
+A `web_search` function tool for general "look it up on the internet"
+questions the other tools do not cover: IGDB is games, the Codex is
+Adeptus Mechanicus lore, the news store is the configured feeds, and
+`fetch_url` needs a URL the user already has. Web search fills the gap
+and pairs with `fetch_url` (search → pick a link → read it). Results are
+external text and are sanitized (SAF-03); the Exa key is a host secret,
+never in the repo. Active only when `enable-web-search = true` and a key
+is present.
+
+### WEB-01 — web_search is offered as a tool (coverage: test)
+
+When `enable-web-search` is true **and** an Exa key is available
+(`exa-api-key` in config, else `EXA_API_KEY` env), the chat call's
+`tools` list includes a `web_search` function (`query` string, optional
+`num_results`). With the flag off or no key it is absent.
+
+### WEB-02 — Results are reduced and sanitized (coverage: test)
+
+Each Exa result becomes `{title, url, snippet, published}`; `title` and
+`snippet` pass through `sanitize_external_text` (snippet capped at
+`web-snippet-chars`, default 400) so a web page can neither inject an
+`@everyone` nor smuggle control characters into the prompt.
+
+### WEB-03 — Result count is bounded (coverage: test)
+
+`num_results` is clamped to 1..`MAX_RESULTS` (10) before the request, so
+neither a huge fan-out nor a zero/negative count reaches the API.
+
+### WEB-04 — Missing key and API failure are reported, not raised (coverage: test)
+
+With no key the tool returns an `{error: ...}` result without a network
+call. A request that raises (network, non-2xx, bad JSON) is logged and
+returns an `{error: ...}` dict — `search` never raises into the loop.
+
+### WEB-05 — Searches are metered per user (coverage: test)
+
+Each `web_search` increments a per-user daily counter; over
+`web-daily-per-user` (default 30) the tool refuses with an error result
+without calling the API. The budget gate (SAF-04) still applies to the
+surrounding model calls.

specs/SPEC-016-weather.md

@@ -0,0 +1,35 @@
+# SPEC-016 — Weather tool (get_weather)
+
+Both personas talk about weather (the sea over the skerries, rain on
+patch day) but had to guess it. `get_weather` grounds that in the
+free MET Norway Locationforecast API (api.met.no, User-Agent
+required, no key). Locations are host-configured coordinates — the
+model picks by name, it never supplies raw URLs, so there is no SSRF
+surface (one fixed API host).
+
+### WEA-01 — Tool offered only when configured (coverage: test)
+
+The chat call's tools include `get_weather` only when
+`enable-weather` is true AND `weather-locations` (a list of
+`[name, lat, lon]` entries) is non-empty. Otherwise it is absent.
+
+### WEA-02 — Compact sanitized forecast (coverage: test)
+
+The tool reduces the MET compact timeseries to: the named location,
+current conditions (temperature °C, wind m/s, symbol), and a small
+set of forecast points (next hours / tomorrow) with temperature,
+symbol and precipitation. Location names pass
+`sanitize_external_text`; numbers are numbers. Nothing else from the
+API response reaches the prompt.
+
+### WEA-03 — Location matched by name, defaults to first (coverage: test)
+
+The `location` argument matches configured entries
+case-insensitively by substring; no or unknown location = the first
+configured entry. Coordinates never come from the model.
+
+### WEA-04 — Errors return, never raise; calls are metered (coverage: test)
+
+API/network failures return an `{error}` dict (the responder keeps
+running). Each call counts against a per-user daily cap
+(`weather-daily-per-user`, default 30) like the other tools.

tests/test_spec_beh.py

@@ -65,6 +65,88 @@ class TestClassifierGate(ClassifierGateBase):
         self.bot.respond.assert_not_awaited()
 
 
+class TestIgnoredChannels(ClassifierGateBase):
+    def ignored_msg(self, channel_name):
+        message = self.public_msg("hello there")
+        message.channel.name = channel_name
+        message.add_reaction = AsyncMock()
+        return message
+
+    async def test_pattern_match_suppresses_reaction_and_reply(self):
+        """BEH-09: fnmatch pattern hit -> no classifier call, no emoji, no reply."""
+        self.gate_setup({"reply": False, "factual": False, "emoji": "👍"})
+        self.bot.config["ignore-channels"] = ["todo*"]
+        message = self.ignored_msg("todo-lists")
+        await self.bot.on_message(message)
+        self.bot.airesponder.classify.assert_not_awaited()
+        message.add_reaction.assert_not_awaited()
+        self.bot.respond.assert_not_awaited()
+
+    async def test_exact_name_still_matches(self):
+        """BEH-09: plain names keep working as exact matches."""
+        self.gate_setup({"reply": True, "factual": False, "emoji": None})
+        self.bot.config["ignore-channels"] = ["blengon"]
+        await self.bot.on_message(self.ignored_msg("blengon"))
+        self.bot.respond.assert_not_awaited()
+
+    async def test_non_matching_channel_passes(self):
+        """BEH-09: unmatched channels reach the responder as before."""
+        self.gate_setup({"reply": True, "factual": False, "emoji": None})
+        self.bot.config["ignore-channels"] = ["todo*"]
+        await self.bot.on_message(self.ignored_msg("chat"))
+        self.bot.respond.assert_awaited_once()
+
+    async def test_dm_never_ignored(self):
+        """BEH-09: a DM whose recipient name matches a pattern is still answered."""
+        self.gate_setup({"reply": True, "factual": False, "emoji": None})
+        self.bot.config["ignore-channels"] = ["todo*"]
+        message = self.public_msg("hei bot")
+        message.channel = MagicMock(spec=DMChannel)
+        message.channel.recipient = MagicMock()
+        message.channel.recipient.name = "todo-fan"
+        await self.bot.on_message(message)
+        self.bot.respond.assert_awaited_once()
+
+    def test_channel_by_name_honors_patterns(self):
+        """BEH-09: channel_by_name resolution skips pattern-ignored channels."""
+        self.bot.config["ignore-channels"] = ["todo*"]
+        fallback = MagicMock(spec=TextChannel)
+        self.assertIs(self.bot.channel_by_name("todo-lists", fallback), fallback)
+
+
+class TestFactualModel(unittest.IsolatedAsyncioTestCase):
+    async def _model_used(self, config, factual):
+        from .test_spec_structured import ok_result
+
+        responder = OpenAIResponder(dict({"openai-token": "t", "model": "cheap", "system": "s", "history-limit": 5}, **config), "chat")
+        responder._factual = factual
+        with patch("fjerkroa_bot.openai_responder.openai_chat", new_callable=AsyncMock) as chat_mock:
+            chat_mock.return_value = ok_result()
+            await responder.chat([{"role": "user", "content": "hi"}], 10)
+        return chat_mock.await_args.kwargs["model"]
+
+    async def test_factual_uses_stronger_model(self):
+        """BEH-10: factual verdict + factual-model config -> stronger tier."""
+        self.assertEqual(await self._model_used({"factual-model": "strong"}, True), "strong")
+
+    async def test_factual_without_config_stays_default(self):
+        """BEH-10: no factual-model config -> default model, no behavior change."""
+        self.assertEqual(await self._model_used({}, True), "cheap")
+
+    async def test_small_talk_stays_default(self):
+        """BEH-10: non-factual messages stay on the cheap default."""
+        self.assertEqual(await self._model_used({"factual-model": "strong"}, False), "cheap")
+
+    async def test_send_reads_flag_from_message(self):
+        """BEH-10: send() picks the factual flag off the AIMessage."""
+        responder = FakeModelResponder({"system": "s", "history-limit": 5}, "chat")
+        responder.scripted = [envelope(answer="x", answer_needed=True)]
+        message = AIMessage("alice", "opening hours?", "chat")
+        message.factual = True
+        await responder.send(message)
+        self.assertTrue(responder._factual)
+
+
 class TestTypingPacing(OpsBase):
     async def send_with(self, answer, factual, cps=30):
         if cps is not None:

tests/test_spec_monitor.py

@@ -0,0 +1,106 @@
+"""Unit coverage for SPEC-012 health monitoring (OPS-18/19)."""
+
+import unittest
+from unittest.mock import AsyncMock
+
+from fjerkroa_bot.monitor import HealthMonitor
+
+
+class FakeLedger:
+    def __init__(self, spent=0.0):
+        self._spent = spent
+
+    def spent_usd(self):
+        return self._spent
+
+
+class FakeStore:
+    def __init__(self, open_tasks=0):
+        self._n = open_tasks
+
+    def tasks_open(self):
+        return list(range(self._n))
+
+
+def _monitor(cfg, ledger=None, store=None, disk=1000.0):
+    alert = AsyncMock()
+    monitor = HealthMonitor(lambda: cfg, ledger or FakeLedger(), store, lambda: disk, alert)
+    return monitor, alert
+
+
+class TestEnabled(unittest.TestCase):
+    def test_opt_in(self):
+        """OPS-18: monitoring is opt-in via enable-monitoring."""
+        self.assertFalse(_monitor({})[0].enabled())
+        self.assertTrue(_monitor({"enable-monitoring": True})[0].enabled())
+
+
+class TestChecks(unittest.IsolatedAsyncioTestCase):
+    async def test_spend_over_threshold_alerts(self):
+        """OPS-18: spend at/above frac*budget alerts."""
+        monitor, alert = _monitor({"daily-budget-usd": 2.0, "monitor-spend-alert-frac": 0.8}, ledger=FakeLedger(1.8))
+        await monitor.tick()
+        alert.assert_awaited_once()
+        self.assertIn("Spend", alert.await_args.args[0])
+
+    async def test_spend_under_threshold_silent(self):
+        """OPS-18: spend below threshold stays silent."""
+        monitor, alert = _monitor({"daily-budget-usd": 2.0}, ledger=FakeLedger(0.5))
+        await monitor.tick()
+        alert.assert_not_awaited()
+
+    async def test_no_budget_skips_spend(self):
+        """OPS-18: no budget configured -> spend check skipped, never fatal."""
+        monitor, alert = _monitor({"enable-monitoring": True}, ledger=FakeLedger(99))
+        await monitor.tick()
+        alert.assert_not_awaited()
+
+    async def test_low_disk_alerts(self):
+        """OPS-18: free disk below the floor alerts."""
+        monitor, alert = _monitor({"monitor-disk-min-mb": 500}, disk=100.0)
+        await monitor.tick()
+        self.assertTrue(any("Low disk" in call.args[0] for call in alert.await_args_list))
+
+    async def test_disk_read_failure_skipped(self):
+        """OPS-18: a failing disk read is skipped, not fatal."""
+
+        def boom():
+            raise OSError("nope")
+
+        alert = AsyncMock()
+        monitor = HealthMonitor(lambda: {}, FakeLedger(), None, boom, alert)
+        await monitor.tick()
+        alert.assert_not_awaited()
+
+    async def test_deep_queue_alerts(self):
+        """OPS-18: task-queue depth at/above max alerts."""
+        monitor, alert = _monitor({"monitor-taskqueue-max": 3}, store=FakeStore(5), disk=9999.0)
+        await monitor.tick()
+        self.assertTrue(any("Task queue" in call.args[0] for call in alert.await_args_list))
+
+    async def test_no_store_skips_queue(self):
+        """OPS-18: no store -> queue check skipped."""
+        monitor, alert = _monitor({"monitor-taskqueue-max": 1}, store=None, disk=9999.0)
+        await monitor.tick()
+        alert.assert_not_awaited()
+
+
+class TestEdgeArming(unittest.IsolatedAsyncioTestCase):
+    async def test_fires_once_per_crossing(self):
+        """OPS-19: a persistent over-threshold condition alerts once, not every tick."""
+        monitor, alert = _monitor({"daily-budget-usd": 2.0}, ledger=FakeLedger(1.9))
+        await monitor.tick()
+        await monitor.tick()
+        await monitor.tick()
+        self.assertEqual(alert.await_count, 1)
+
+    async def test_rearms_on_recovery(self):
+        """OPS-19: recovery re-arms silently; the next crossing alerts again."""
+        ledger = FakeLedger(1.9)
+        monitor, alert = _monitor({"daily-budget-usd": 2.0}, ledger=ledger, disk=9999.0)
+        await monitor.tick()  # over -> alert (1)
+        ledger._spent = 0.5
+        await monitor.tick()  # recovered -> silent, re-arm
+        ledger._spent = 1.95
+        await monitor.tick()  # over again -> alert (2)
+        self.assertEqual(alert.await_count, 2)

tests/test_spec_news.py

@@ -1,9 +1,24 @@
-"""Unit coverage for SPEC-013 news digest (NEWS-01..03)."""
+"""Unit coverage for SPEC-013 news digest + memory + tool (NEWS-01..12)."""
 
+import tempfile
 import unittest
+from pathlib import Path
 from unittest.mock import AsyncMock
 
-from fjerkroa_bot.news import NewsFetcher, NewsPoster, load_seen, parse_feed, render_digest, save_seen
+from fjerkroa_bot.news import (
+    GET_NEWS_TOOL,
+    NewsFetcher,
+    NewsPoster,
+    load_seen,
+    parse_feed,
+    query_news,
+    render_digest,
+    save_seen,
+)
+from fjerkroa_bot.openai_responder import OpenAIResponder
+from fjerkroa_bot.persistence import PersistentStore
+
+CONFIG = {"openai-token": "t", "model": "m", "system": "s", "history-limit": 5}
 
 RSS = b"""<?xml version="1.0"?><rss><channel>
 <item><title>Game X released</title><link>https://ex.com/x</link></item>
@@ -14,6 +29,16 @@ ATOM = b"""<?xml version="1.0"?><feed xmlns="http://www.w3.org/2005/Atom">
 <entry><title>Atom headline</title><link href="https://ex.com/a"/></entry>
 </feed>"""
 
+RSS1 = (
+    '<?xml version="1.0" encoding="UTF-8"?>'
+    '<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://purl.org/rss/1.0/">'
+    '<channel rdf:about="https://ex.jp"><title>Feed</title></channel>'
+    '<item rdf:about="https://ex.jp/1"><title>ゲームニュース</title><link>https://ex.jp/1</link>'
+    "<description>本文ここ</description></item>"
+    '<item rdf:about="https://ex.jp/2"><title>Second</title><link>https://ex.jp/2</link></item>'
+    "</rdf:RDF>"
+).encode("utf-8")
+
 
 class TestParse(unittest.TestCase):
     def test_rss(self):
@@ -29,6 +54,13 @@ class TestParse(unittest.TestCase):
         self.assertEqual(items[0]["title"], "Atom headline")
         self.assertEqual(items[0]["link"], "https://ex.com/a")
 
+    def test_rss1_rdf(self):
+        """NEWS-01: RSS 1.0/RDF (namespaced <item>, e.g. 4gamer.net) parses like RSS 2.0."""
+        items = parse_feed(RSS1, "JP")
+        self.assertEqual([i["title"] for i in items], ["ゲームニュース", "Second"])
+        self.assertEqual(items[0]["link"], "https://ex.jp/1")
+        self.assertEqual(items[0]["summary"], "本文ここ")
+
     def test_malformed_never_raises(self):
         """NEWS-01: garbage XML returns [] without raising."""
         self.assertEqual(parse_feed(b"<not xml", "bad"), [])
@@ -166,10 +198,157 @@ class TestSeenState(unittest.TestCase):
     def test_cap_bounds_state(self):
         """NEWS-05: save keeps at most `cap` keys."""
         import json
-        import tempfile
-        from pathlib import Path
 
         with tempfile.TemporaryDirectory() as tmp:
             path = str(Path(tmp) / "state.json")
             save_seen(path, {f"k{i}" for i in range(100)}, cap=10)
             self.assertEqual(len(json.load(open(path))), 10)
+
+
+RSS_DESC = b"""<?xml version="1.0"?><rss><channel>
+<item><title>Storm hits coast</title><link>https://ex.com/s</link>
+<description>&lt;p&gt;Heavy &lt;b&gt;wind&lt;/b&gt;   expected&lt;/p&gt;</description></item>
+</channel></rss>"""
+
+ATOM_SUM = b"""<?xml version="1.0"?><feed xmlns="http://www.w3.org/2005/Atom">
+<entry><title>Atom T</title><link href="https://ex.com/a"/><summary>Short  gist  here</summary></entry>
+</feed>"""
+
+
+class TestSummaries(unittest.TestCase):
+    def test_rss_description_stripped(self):
+        """NEWS-07: RSS description parsed, HTML stripped, entities unescaped, whitespace collapsed."""
+        items = parse_feed(RSS_DESC, "S")
+        self.assertEqual(items[0]["summary"], "Heavy wind expected")
+
+    def test_atom_summary(self):
+        """NEWS-07: Atom summary collapsed to clean text."""
+        items = parse_feed(ATOM_SUM, "A")
+        self.assertEqual(items[0]["summary"], "Short gist here")
+
+    def test_missing_description_is_empty(self):
+        """NEWS-07: no description -> empty summary, never an error."""
+        self.assertEqual(parse_feed(RSS, "S")[0]["summary"], "")
+
+    def test_digest_carries_summary(self):
+        """NEWS-08: digest appends the sanitized capped summary; zero cap = title only."""
+        items = [{"title": "T", "link": "https://ex.com/x", "source": "NRK", "summary": "the gist of it"}]
+        digest = render_digest(items, 10, 100)
+        self.assertIn("[NRK]", digest)
+        self.assertIn("the gist of it", digest)
+        self.assertNotIn("the gist", render_digest(items, 10, 0))  # zero cap -> title only
+
+
+class NewsStoreBase(unittest.TestCase):
+    def setUp(self):
+        self.tmp = tempfile.TemporaryDirectory()
+        self.addCleanup(self.tmp.cleanup)
+        self.store = PersistentStore(Path(self.tmp.name) / "bot.db")
+
+
+class TestNewsStore(NewsStoreBase):
+    def test_dedup_and_rolling_window(self):
+        """NEWS-09: items deduped by link; prune keeps the newest N."""
+        first = [
+            {"title": "A", "link": "L1", "source": "S", "summary": "sa"},
+            {"title": "B", "link": "L2", "source": "S", "summary": "sb"},
+        ]
+        self.assertEqual(self.store.add_news_items(first), 2)
+        self.assertEqual(self.store.add_news_items([dict(first[0])]), 0)  # dup link ignored
+        self.assertEqual(self.store.news_count(), 2)
+        self.store.prune_news(1)
+        self.assertEqual(self.store.news_count(), 1)
+        self.assertEqual(self.store.recent_news(5)[0]["title"], "B")  # newest survives
+
+    def test_dedup_by_title_when_no_link(self):
+        """NEWS-09: linkless items dedup on title."""
+        self.store.add_news_items([{"title": "Same", "link": "", "source": "S", "summary": ""}])
+        self.store.add_news_items([{"title": "Same", "link": "", "source": "S", "summary": ""}])
+        self.assertEqual(self.store.news_count(), 1)
+
+
+class TestQueryNews(NewsStoreBase):
+    def seed(self):
+        self.store.add_news_items(
+            [
+                {"title": "Nordland storm", "link": "L1", "source": "Nordland", "summary": "strong wind on the coast"},
+                {"title": "Oslo budget", "link": "L2", "source": "NRK", "summary": "@everyone spending plan"},
+                {"title": "Sport result", "link": "L3", "source": "Sport", "summary": "the match ended"},
+            ]
+        )
+
+    def test_topic_filter(self):
+        """NEWS-11: topic keywords must appear in title or summary."""
+        self.seed()
+        res = query_news(self.store, topic="storm")
+        self.assertEqual([r["title"] for r in res["results"]], ["Nordland storm"])
+
+    def test_topic_matches_source_label(self):
+        """NEWS-11: topic also matches the source label, so 'Nordland' finds regional items."""
+        self.store.add_news_items([{"title": "Ferry delayed", "link": "LX", "source": "Nordland", "summary": "boat late"}])
+        res = query_news(self.store, topic="Nordland")
+        self.assertTrue(any(r["link"] == "LX" for r in res["results"]))  # matched via source, not title/summary
+
+    def test_source_filter_and_sanitize(self):
+        """NEWS-11: source narrows results; title/summary are sanitized."""
+        self.seed()
+        res = query_news(self.store, source="NRK")
+        self.assertTrue(res["results"] and all(r["source"] == "NRK" for r in res["results"]))
+        self.assertNotIn("@everyone", res["results"][0]["summary"])
+
+    def test_limit_clamped_and_no_store(self):
+        """NEWS-11: limit clamps to 1..30; a missing store returns an error."""
+        self.seed()
+        self.assertLessEqual(len(query_news(self.store, limit=999)["results"]), 30)
+        self.assertGreaterEqual(len(query_news(self.store, limit=0)["results"]), 1)
+        self.assertIn("error", query_news(None))
+
+    def test_exact_match_has_no_note(self):
+        """NEWS-13: a direct AND-match returns without a note field."""
+        self.seed()
+        self.assertNotIn("note", query_news(self.store, topic="storm"))
+
+    def test_partial_match_falls_back_ranked(self):
+        """NEWS-13: AND-miss -> any-term match, most keyword hits first, with a note."""
+        self.seed()
+        res = query_news(self.store, topic="Nordland road accident")
+        self.assertEqual(res["results"][0]["title"], "Nordland storm")
+        self.assertIn("note", res)
+
+    def test_no_match_falls_back_to_recent(self):
+        """NEWS-13: nothing matches any term -> newest items + note, never empty-handed."""
+        self.seed()
+        res = query_news(self.store, topic="quantum blockchain")
+        self.assertTrue(res["results"])
+        self.assertEqual(res["results"][0]["title"], "Sport result")  # newest first
+        self.assertIn("note", res)
+
+
+class TestNewsTool(unittest.IsolatedAsyncioTestCase):
+    def setUp(self):
+        self.tmp = tempfile.TemporaryDirectory()
+        self.addCleanup(self.tmp.cleanup)
+
+    def _responder(self, **extra):
+        cfg = dict(CONFIG, **{"history-directory": self.tmp.name}, **extra)
+        return OpenAIResponder(cfg, "chat")
+
+    def test_tool_offered_needs_flag_and_store(self):
+        """NEWS-10: get_news offered only with enable-news-tool AND a store."""
+        no_store = OpenAIResponder(dict(CONFIG, **{"enable-news-tool": True}), "chat")
+        self.assertIsNone(no_store.store)
+        self.assertNotIn("get_news", [f["name"] for f in no_store._available_tools()])
+        flag_off = self._responder()
+        self.assertNotIn("get_news", [f["name"] for f in flag_off._available_tools()])
+        on = self._responder(**{"enable-news-tool": True})
+        self.assertIn("get_news", [f["name"] for f in on._available_tools()])
+        self.assertEqual(GET_NEWS_TOOL["name"], "get_news")
+
+    async def test_dispatch_caps_news(self):
+        """NEWS-12: over news-daily-per-user, get_news refuses without querying."""
+        responder = self._responder(**{"enable-news-tool": True, "news-daily-per-user": 2})
+        responder.store.add_news_items([{"title": "x", "link": "l", "source": "s", "summary": "y"}])
+        for _ in range(2):
+            self.assertIn("results", await responder._dispatch_tool("get_news", {}, "alice"))
+        blocked = await responder._dispatch_tool("get_news", {}, "alice")
+        self.assertIn("error", blocked)

tests/test_spec_ops.py

@@ -133,3 +133,69 @@ class TestTasksKillSwitch(OpsBase):
         """OPS-09: bot-initiated posts respect pause/quiet."""
         await self.bot.on_message(self.staff_msg("!bot pause"))
         self.assertFalse(self.bot.bot_initiated_allowed())
+
+
+class TestHelp(OpsBase):
+    STAFF_CMDS = (
+        "pause",
+        "resume",
+        "quiet <minutes>",
+        "status",
+        "spend",
+        "images on|off",
+        "memory <user>",
+        "forget-fact <id>",
+        "pin <channel|global>",
+        "unpin <id>",
+        "pins",
+        "task-approve <id>",
+        "task-cancel <id>",
+        "(list)",
+    )
+
+    def test_staff_help_is_complete_and_grouped(self):
+        """OPS-17: staff help lists every operator command, grouped by purpose."""
+        text = self.bot._help_text(staff=True)
+        for cmd in self.STAFF_CMDS:
+            self.assertIn(cmd, text, f"missing {cmd!r} in staff help")
+        for group in ("Control:", "Cost:", "Memory:", "Tasks:"):
+            self.assertIn(group, text)
+        for cmd in ("!help", "!forgetme", "!privacy", "!wichtel"):
+            self.assertIn(cmd, text)  # everywhere-commands shown too
+
+    def test_user_help_hides_operator_commands(self):
+        """OPS-17: non-staff help shows only the everyone-commands."""
+        text = self.bot._help_text(staff=False)
+        for cmd in ("!help", "!forgetme", "!privacy", "!wichtel"):
+            self.assertIn(cmd, text)
+        for op in ("task-approve", "images on|off", "spend", "Staff commands", "Control:"):
+            self.assertNotIn(op, text)
+
+    async def test_bot_help_in_staff_channel_returns_full_help(self):
+        """OPS-17: `!bot help` answers with the complete staff help."""
+        await self.bot.on_message(self.staff_msg("!bot help"))
+        text = self.bot.staff_channel.send.await_args.args[0]
+        self.assertIn("Staff commands", text)
+        self.assertIn("task-cancel <id>", text)
+
+    async def test_unknown_bot_command_falls_back_to_help(self):
+        """OPS-17: an unrecognised `!bot` command shows the full help, not a partial line."""
+        await self.bot.on_message(self.staff_msg("!bot wat"))
+        text = self.bot.staff_channel.send.await_args.args[0]
+        self.assertIn("Control:", text)
+
+    async def test_help_in_public_channel_is_user_scoped(self):
+        """OPS-17: `!help` in a normal channel lists only everyone-commands."""
+        msg = self.public_msg("!help")
+        await self.bot.on_message(msg)
+        text = msg.channel.send.await_args.args[0]
+        self.assertIn("!forgetme", text)
+        self.assertNotIn("Staff commands", text)
+        self.assertNotIn("task-approve", text)
+
+    async def test_help_works_while_paused(self):
+        """OPS-17: help answers even when replies are paused."""
+        await self.bot.on_message(self.staff_msg("!bot pause"))
+        msg = self.public_msg("!help")
+        await self.bot.on_message(msg)
+        msg.channel.send.assert_awaited()

tests/test_spec_per.py

@@ -120,9 +120,7 @@ class TestConfigReloadRace(TestBotBase):
         new_config["history-limit"] = 99
         self.bot.load_config = lambda path: new_config
         self.bot.loop = MagicMock()
-        event = MagicMock()
-        event.src_path = self.bot.config_file
-        self.bot.on_config_file_modified(event)
+        self.bot.on_config_file_changed()
         self.bot.loop.call_soon_threadsafe.assert_called_once()
         apply_fn = self.bot.loop.call_soon_threadsafe.call_args.args[0]
         apply_fn()
@@ -137,7 +135,37 @@ class TestConfigReloadRace(TestBotBase):
         loop = MagicMock()
         loop.call_soon_threadsafe.side_effect = RuntimeError("no running loop")
         self.bot.loop = loop
-        event = MagicMock()
-        event.src_path = self.bot.config_file
-        self.bot.on_config_file_modified(event)
+        self.bot.on_config_file_changed()
         self.assertEqual(self.bot.config["history-limit"], 42)
+
+
+class TestConfigReloadRenameSafe(unittest.TestCase):
+    def test_atomic_rename_and_modify_trigger_reload(self):
+        """CFG-05: a modified OR a renamed-into-place config fires the reload; unrelated files do not."""
+        from fjerkroa_bot.discord_bot import ConfigFileHandler
+
+        with tempfile.TemporaryDirectory() as tmp:
+            config = Path(tmp) / "kroa.toml"
+            config.write_text("x = 1\n")
+            hits = []
+            handler = ConfigFileHandler(str(config), lambda: hits.append(1))
+
+            def evt(is_dir=False, src=None, dest=None):
+                event = MagicMock()
+                event.is_directory = is_dir
+                event.src_path = src if src is not None else ""
+                event.dest_path = dest if dest is not None else ""
+                return event
+
+            handler.on_modified(evt(src=str(config)))  # in-place modify
+            handler.on_moved(evt(src=str(Path(tmp) / "kroa.toml.tmp"), dest=str(config)))  # atomic rename over
+            handler.on_created(evt(src=str(config)))  # write-new
+            self.assertEqual(len(hits), 3)
+
+            handler.on_modified(evt(src=str(Path(tmp) / "other.txt")))  # unrelated file
+            handler.on_modified(evt(is_dir=True, src=str(config)))  # directory event
+            self.assertEqual(len(hits), 3)  # neither fired
+
+            # open/close of the config (our own load_config re-reads) must NOT be handled — else a reload loop.
+            self.assertNotIn("on_opened", vars(ConfigFileHandler))
+            self.assertNotIn("on_closed", vars(ConfigFileHandler))

tests/test_spec_responses.py

@@ -0,0 +1,165 @@
+"""Unit coverage for the Responses API path (ENV-22..24, D-021)."""
+
+import json
+import unittest
+from unittest.mock import AsyncMock, Mock, patch
+
+from fjerkroa_bot.openai_responder import ENVELOPE_TEXT_FORMAT, OpenAIResponder
+
+from .test_bdd_envelope import envelope
+
+CONFIG = {
+    "openai-token": "t",
+    "model": "main-model",
+    "system": "s",
+    "history-limit": 5,
+    "use-responses-api": True,
+    "reasoning-effort": "medium",
+}
+
+
+def _msg_item():
+    part = Mock()
+    part.type = "output_text"
+    item = Mock()
+    item.type = "message"
+    item.content = [part]
+    item.model_dump = lambda: {"type": "message"}
+    return item
+
+
+def _refusal_item():
+    part = Mock()
+    part.type = "refusal"
+    item = Mock()
+    item.type = "message"
+    item.content = [part]
+    return item
+
+
+def _reasoning_item():
+    item = Mock()
+    item.type = "reasoning"
+    item.id = "rs_1"
+    item.summary = []
+    item.encrypted_content = "opaque-cot"
+    item.status = "completed"  # response-only field; must NOT travel back
+    return item
+
+
+def _call_item(name, args, call_id="call-1"):
+    item = Mock()
+    item.type = "function_call"
+    item.id = "fc_1"
+    item.name = name
+    item.arguments = json.dumps(args)
+    item.call_id = call_id
+    item.status = "completed"
+    return item
+
+
+def _response(output, text=""):
+    result = Mock()
+    result.output = output
+    result.output_text = text
+    result.usage = Mock(prompt_tokens=None, completion_tokens=None, input_tokens=5, output_tokens=7)
+    return result
+
+
+class TestResponsesPath(unittest.IsolatedAsyncioTestCase):
+    def _responder(self, **extra):
+        return OpenAIResponder(dict(CONFIG, **extra), "chat")
+
+    async def test_flag_routes_to_responses_with_reasoning(self):
+        """ENV-22: flag on -> /v1/responses with envelope text.format, reasoning from config, stateless kwargs."""
+        responder = self._responder()
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            with patch("fjerkroa_bot.openai_responder.openai_chat", new_callable=AsyncMock) as chat_mock:
+                responses_mock.return_value = _response([_msg_item()], envelope(answer="hi", answer_needed=True))
+                answer, _ = await responder.chat([{"role": "user", "content": "hei"}], 10)
+        chat_mock.assert_not_awaited()
+        self.assertEqual(json.loads(answer["content"])["answer"], "hi")
+        kwargs = responses_mock.await_args.kwargs
+        self.assertEqual(kwargs["text"], ENVELOPE_TEXT_FORMAT)
+        self.assertEqual(kwargs["reasoning"], {"effort": "medium"})
+        self.assertFalse(kwargs["store"])  # ENV-23
+        self.assertIn("reasoning.encrypted_content", kwargs["include"])
+
+    async def test_flag_off_stays_on_chat_completions(self):
+        """ENV-22: flag off (default) -> openai_responses never called."""
+        from .test_spec_structured import ok_result
+
+        responder = OpenAIResponder({k: v for k, v in CONFIG.items() if k != "use-responses-api"}, "chat")
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            with patch("fjerkroa_bot.openai_responder.openai_chat", new_callable=AsyncMock) as chat_mock:
+                chat_mock.return_value = ok_result()
+                await responder.chat([{"role": "user", "content": "hei"}], 10)
+        responses_mock.assert_not_awaited()
+        chat_mock.assert_awaited()
+
+    async def test_tools_flat_shape(self):
+        """ENV-22: tools are sent in the flat Responses shape (name at top level)."""
+        responder = self._responder(**{"enable-news-tool": True})
+        responder.store = Mock()  # store present -> get_news offered
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            responses_mock.return_value = _response([_msg_item()], envelope(answer="x", answer_needed=True))
+            await responder.chat([{"role": "user", "content": "hei"}], 10)
+        tools = responses_mock.await_args.kwargs["tools"]
+        self.assertTrue(all(tool["type"] == "function" and "name" in tool and "function" not in tool for tool in tools))
+
+    async def test_tool_loop_passes_reasoning_and_outputs_back(self):
+        """ENV-23: function_call -> dispatch; next call carries reasoning item + function_call_output."""
+        responder = self._responder(**{"enable-news-tool": True})
+        responder.store = Mock()
+        responder._dispatch_tool = AsyncMock(return_value={"results": ["ok"]})
+        first = _response([_reasoning_item(), _call_item("get_news", {"topic": "x"}, "call-9")])
+        second = _response([_msg_item()], envelope(answer="done", answer_needed=True))
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            responses_mock.side_effect = [first, second]
+            answer, _ = await responder.chat([{"role": "user", "content": "news?"}], 10)
+        self.assertEqual(json.loads(answer["content"])["answer"], "done")
+        responder._dispatch_tool.assert_awaited_once()
+        followup_input = responses_mock.await_args_list[1].kwargs["input"]
+        reasoning = [item for item in followup_input if isinstance(item, dict) and item.get("type") == "reasoning"]
+        self.assertEqual(len(reasoning), 1)
+        self.assertEqual(reasoning[0]["encrypted_content"], "opaque-cot")
+        self.assertNotIn("status", reasoning[0])  # response-only field stripped (live-400 regression)
+        calls_back = [item for item in followup_input if isinstance(item, dict) and item.get("type") == "function_call"]
+        self.assertNotIn("status", calls_back[0])
+        outputs = [item for item in followup_input if isinstance(item, dict) and item.get("type") == "function_call_output"]
+        self.assertEqual(len(outputs), 1)
+        self.assertEqual(outputs[0]["call_id"], "call-9")
+
+    async def test_exhausted_rounds_force_toolless_answer(self):
+        """ENV-23: after responses-tool-rounds rounds the final call drops tools."""
+        responder = self._responder(**{"enable-news-tool": True, "responses-tool-rounds": 1})
+        responder.store = Mock()
+        responder._dispatch_tool = AsyncMock(return_value={"results": []})
+        looping = _response([_call_item("get_news", {}, "c")])
+        final = _response([_msg_item()], envelope(answer="forced", answer_needed=True))
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            responses_mock.side_effect = [looping, final]
+            answer, _ = await responder.chat([{"role": "user", "content": "go"}], 10)
+        self.assertEqual(json.loads(answer["content"])["answer"], "forced")
+        self.assertNotIn("tools", responses_mock.await_args_list[1].kwargs)
+
+    async def test_refusal_is_failed_attempt(self):
+        """ENV-24: a refusal part -> no answer."""
+        responder = self._responder()
+        with patch("fjerkroa_bot.openai_responder.openai_responses", new_callable=AsyncMock) as responses_mock:
+            responses_mock.return_value = _response([_refusal_item()])
+            answer, _ = await responder.chat([{"role": "user", "content": "hei"}], 10)
+        self.assertIsNone(answer)
+
+    async def test_vision_parts_mapped(self):
+        """ENV-22: chat-format image parts become input_image items."""
+        items = OpenAIResponder._responses_input(
+            [
+                {"role": "user", "content": [{"type": "text", "text": "look"}, {"type": "image_url", "image_url": {"url": "data:x"}}]},
+                {"role": "tool", "content": "dropped"},
+                {"role": "assistant", "content": "{}"},
+            ]
+        )
+        self.assertEqual(items[0]["content"][0], {"type": "input_text", "text": "look"})
+        self.assertEqual(items[0]["content"][1], {"type": "input_image", "image_url": "data:x"})
+        self.assertEqual(len(items), 2)  # tool row dropped

tests/test_spec_url.py

@@ -149,6 +149,31 @@ class TestTextExtraction(unittest.TestCase):
         self.assertNotIn("evil", text)
         self.assertNotIn("x{}", text)
 
+    def test_chrome_and_link_boilerplate_dropped(self):
+        """URL-08: nav/header/footer skipped; short link-dominated blocks (menus, related lists) removed."""
+        reader = URLReader(lambda: {}, None)
+        html = (
+            "<html><body>"
+            "<nav><a href='/a'>Home</a> <a href='/b'>Games</a></nav>"
+            "<header><a href='/login'>Login</a></header>"
+            "<ul><li><a href='/1'>Related article one</a></li><li><a href='/2'>Related article two</a></li></ul>"
+            "<article><p>The pop-up event runs from August 4 in Shibuya, with details "
+            "<a href='/x'>on the official page</a> for anyone attending the exhibition.</p></article>"
+            "<footer><a href='/imprint'>Imprint</a></footer>"
+            "</body></html>"
+        )
+        text = reader._to_text(html)
+        self.assertIn("pop-up event", text)
+        self.assertIn("on the official page", text)  # inline link in a real paragraph survives
+        for chrome in ("Home", "Login", "Related article one", "Imprint"):
+            self.assertNotIn(chrome, text)
+
+    def test_default_cap_is_8000(self):
+        """URL-08: the default url-max-chars budget is 8000."""
+        from fjerkroa_bot.url_reader import DEFAULT_MAX_CHARS
+
+        self.assertEqual(DEFAULT_MAX_CHARS, 8000)
+
 
 class TestBodyReadCollectsAllChunks(unittest.IsolatedAsyncioTestCase):
     async def test_get_reads_past_first_chunk(self):

tests/test_spec_weather.py

@@ -0,0 +1,120 @@
+"""Unit coverage for SPEC-016 weather tool (WEA-01..04)."""
+
+import unittest
+from unittest.mock import AsyncMock, patch
+
+from fjerkroa_bot.openai_responder import OpenAIResponder
+from fjerkroa_bot.weather import GET_WEATHER_TOOL, Weather, _reduce
+
+CONFIG = {"openai-token": "t", "model": "m", "system": "s", "history-limit": 5}
+LOCATIONS = [["Sleneset", 66.58, 12.68], ["Berlin", 52.52, 13.41]]
+
+MET_DATA = {
+    "properties": {
+        "timeseries": [
+            {
+                "time": f"2026-07-17T{10 + i if 10 + i < 24 else 10 + i - 24:02d}:00:00Z",
+                "data": {
+                    "instant": {"details": {"air_temperature": 14.0 + i, "wind_speed": 5.0}},
+                    "next_1_hours": {"summary": {"symbol_code": "lightrain"}, "details": {"precipitation_amount": 0.3}},
+                },
+            }
+            for i in range(30)
+        ]
+    }
+}
+
+
+def _tool_names(responder):
+    return [f["name"] for f in responder._available_tools()]
+
+
+class TestToolOffered(unittest.TestCase):
+    def test_gate_needs_flag_and_locations(self):
+        """WEA-01: get_weather offered only with enable-weather AND locations."""
+        self.assertNotIn("get_weather", _tool_names(OpenAIResponder(CONFIG, "chat")))
+        flag_only = OpenAIResponder(dict(CONFIG, **{"enable-weather": True}), "chat")
+        self.assertNotIn("get_weather", _tool_names(flag_only))
+        on = OpenAIResponder(dict(CONFIG, **{"enable-weather": True, "weather-locations": LOCATIONS}), "chat")
+        self.assertIn("get_weather", _tool_names(on))
+        self.assertEqual(GET_WEATHER_TOOL["name"], "get_weather")
+
+
+class TestReduce(unittest.TestCase):
+    def test_compact_shape(self):
+        """WEA-02: now + few forecast points; temperature/wind/conditions/precip only."""
+        out = _reduce(MET_DATA, "Sleneset")
+        self.assertEqual(out["location"], "Sleneset")
+        self.assertEqual(out["now"]["temp_c"], 14.0)
+        self.assertEqual(out["now"]["wind_ms"], 5.0)
+        self.assertEqual(out["now"]["conditions"], "lightrain")
+        self.assertEqual(out["now"]["precip_mm"], 0.3)
+        self.assertEqual(len(out["forecast"]), 3)  # +6h, +12h, +24h
+        self.assertEqual(out["forecast"][0]["temp_c"], 20.0)
+        self.assertNotIn("error", out)
+
+    def test_location_name_sanitized_and_empty_series(self):
+        """WEA-02: name passes sanitizer; empty timeseries -> error dict."""
+        out = _reduce(MET_DATA, "@everyone town")
+        self.assertNotIn("@everyone", out["location"])
+        self.assertIn("error", _reduce({"properties": {"timeseries": []}}, "x"))
+
+
+class TestLocationPick(unittest.TestCase):
+    def setUp(self):
+        self.weather = Weather(lambda: {"enable-weather": True, "weather-locations": LOCATIONS})
+
+    def test_substring_case_insensitive(self):
+        """WEA-03: case-insensitive substring match."""
+        self.assertEqual(self.weather._pick("berlin")[0], "Berlin")
+        self.assertEqual(self.weather._pick("slen")[0], "Sleneset")
+
+    def test_unknown_or_absent_defaults_to_first(self):
+        """WEA-03: unknown/absent location -> first configured entry."""
+        self.assertEqual(self.weather._pick(None)[0], "Sleneset")
+        self.assertEqual(self.weather._pick("Atlantis")[0], "Sleneset")
+
+    def test_bad_entries_skipped(self):
+        """WEA-03: malformed location entries are ignored, not fatal."""
+        weather = Weather(lambda: {"enable-weather": True, "weather-locations": [["broken"], ["OK", 1.0, 2.0]]})
+        self.assertEqual(weather._pick(None)[0], "OK")
+
+
+class TestForecast(unittest.IsolatedAsyncioTestCase):
+    async def test_error_returned_not_raised(self):
+        """WEA-04: network failure -> {error}, never an exception."""
+        weather = Weather(lambda: {"enable-weather": True, "weather-locations": LOCATIONS})
+        with patch.object(Weather, "_fetch_json", new_callable=AsyncMock, side_effect=RuntimeError("boom")):
+            out = await weather.forecast("Berlin")
+        self.assertIn("error", out)
+
+    async def test_forecast_happy_path(self):
+        """WEA-02/03: full flow with mocked API."""
+        weather = Weather(lambda: {"enable-weather": True, "weather-locations": LOCATIONS})
+        with patch.object(Weather, "_fetch_json", new_callable=AsyncMock, return_value=MET_DATA):
+            out = await weather.forecast("berlin")
+        self.assertEqual(out["location"], "Berlin")
+        self.assertEqual(out["now"]["temp_c"], 14.0)
+
+
+class TestMetering(unittest.IsolatedAsyncioTestCase):
+    async def test_daily_cap(self):
+        """WEA-04: per-user daily cap refuses beyond weather-daily-per-user."""
+        import tempfile
+
+        with tempfile.TemporaryDirectory() as tmp:
+            config = dict(
+                CONFIG,
+                **{
+                    "enable-weather": True,
+                    "weather-locations": LOCATIONS,
+                    "weather-daily-per-user": 1,
+                    "history-directory": tmp,
+                },
+            )
+            responder = OpenAIResponder(config, "chat")
+            with patch.object(Weather, "_fetch_json", new_callable=AsyncMock, return_value=MET_DATA):
+                first = await responder._dispatch_tool("get_weather", {}, "alice")
+                second = await responder._dispatch_tool("get_weather", {}, "alice")
+        self.assertNotIn("error", first)
+        self.assertIn("error", second)

tests/test_spec_websearch.py

@@ -0,0 +1,102 @@
+"""Unit coverage for SPEC-015 web search via Exa (WEB-01..05)."""
+
+import os
+import unittest
+from unittest.mock import AsyncMock, patch
+
+from fjerkroa_bot.openai_responder import OpenAIResponder
+from fjerkroa_bot.websearch import WEB_SEARCH_TOOL, WebSearch, _format_results
+
+CONFIG = {"openai-token": "t", "model": "m", "system": "s", "history-limit": 5}
+
+
+def _tool_names(responder):
+    return [f["name"] for f in responder._available_tools()]
+
+
+class TestToolOffered(unittest.TestCase):
+    def test_gate_needs_flag_and_key(self):
+        """WEB-01: web_search offered only with enable-web-search AND a key."""
+        off = OpenAIResponder(CONFIG, "chat")  # flag off -> absent even if env key exists
+        self.assertNotIn("web_search", _tool_names(off))
+        on = OpenAIResponder(dict(CONFIG, **{"enable-web-search": True, "exa-api-key": "k"}), "chat")
+        self.assertIn("web_search", _tool_names(on))
+        self.assertEqual(WEB_SEARCH_TOOL["name"], "web_search")
+        with patch.dict(os.environ, {"EXA_API_KEY": ""}):
+            nokey = OpenAIResponder(dict(CONFIG, **{"enable-web-search": True}), "chat")
+            self.assertNotIn("web_search", _tool_names(nokey))
+
+
+class TestFormat(unittest.TestCase):
+    def test_results_sanitized_and_capped(self):
+        """WEB-02: title/snippet sanitized + capped; non-dict rows skipped."""
+        data = {
+            "results": [
+                {"title": "@everyone Hi", "url": "https://x.com/a", "text": "@here " + "y" * 1000, "publishedDate": "2026-01-01"},
+                {"title": "T2", "url": "https://x.com/b", "text": "short"},
+                "not a dict",
+            ]
+        }
+        rows = _format_results(data, 50)
+        self.assertEqual(len(rows), 2)
+        self.assertNotIn("@everyone", rows[0]["title"])
+        self.assertNotIn("@here", rows[0]["snippet"])
+        self.assertLessEqual(len(rows[0]["snippet"]), 50)
+        self.assertEqual(rows[0]["url"], "https://x.com/a")
+        self.assertEqual(rows[0]["published"], "2026-01-01")
+
+
+class TestSearch(unittest.IsolatedAsyncioTestCase):
+    async def test_num_results_clamped(self):
+        """WEB-03: numResults clamped to 1..10; 0 falls back to default."""
+        ws = WebSearch(lambda: {"exa-api-key": "k"})
+        with patch.object(ws, "_post", new=AsyncMock(return_value={"results": []})) as post:
+            await ws.search("hi", num_results=999)
+            self.assertEqual(post.await_args.args[0]["numResults"], 10)
+            await ws.search("hi", num_results=0)
+            self.assertEqual(post.await_args.args[0]["numResults"], 5)
+
+    async def test_no_key_returns_error(self):
+        """WEB-04: no key -> error dict, no network call."""
+        with patch.dict(os.environ, {"EXA_API_KEY": ""}):
+            ws = WebSearch(lambda: {})
+            with patch.object(ws, "_post", new=AsyncMock()) as post:
+                result = await ws.search("hi")
+            post.assert_not_awaited()
+        self.assertIn("error", result)
+
+    async def test_api_failure_returns_error(self):
+        """WEB-04: a raising request is caught, returns an error dict."""
+        ws = WebSearch(lambda: {"exa-api-key": "k"})
+        with patch.object(ws, "_post", new=AsyncMock(side_effect=RuntimeError("boom"))):
+            result = await ws.search("hi")
+        self.assertIn("error", result)
+
+    async def test_empty_query_no_call(self):
+        """WEB-04: blank query returns empty results without a call."""
+        ws = WebSearch(lambda: {"exa-api-key": "k"})
+        with patch.object(ws, "_post", new=AsyncMock()) as post:
+            result = await ws.search("   ")
+        post.assert_not_awaited()
+        self.assertEqual(result["results"], [])
+
+    async def test_search_returns_formatted(self):
+        """WEB-02: a successful search returns sanitized rows."""
+        ws = WebSearch(lambda: {"exa-api-key": "k"})
+        payload = {"results": [{"title": "Norge", "url": "https://ex.com/n", "text": "fakta"}]}
+        with patch.object(ws, "_post", new=AsyncMock(return_value=payload)):
+            result = await ws.search("norge")
+        self.assertEqual(result["results"][0]["title"], "Norge")
+        self.assertEqual(result["results"][0]["url"], "https://ex.com/n")
+
+
+class TestPerUserCap(unittest.IsolatedAsyncioTestCase):
+    async def test_dispatch_caps_searches(self):
+        """WEB-05: over web-daily-per-user, web_search refuses without calling the API."""
+        responder = OpenAIResponder(dict(CONFIG, **{"enable-web-search": True, "exa-api-key": "k", "web-daily-per-user": 2}), "chat")
+        responder.web_search.search = AsyncMock(return_value={"query": "x", "results": []})
+        for _ in range(2):
+            self.assertIn("results", await responder._dispatch_tool("web_search", {"query": "hi"}, "bob"))
+        blocked = await responder._dispatch_tool("web_search", {"query": "hi"}, "bob")
+        self.assertIn("error", blocked)
+        self.assertEqual(responder.web_search.search.await_count, 2)