human behavior: classifier gate, pacing, splitting, quiet hours, stable prompt prefix

specs/SPEC-001-responder-envelope.md

@@ -66,13 +66,23 @@ link syntax from the model reads as noise.
 When the envelope `channel` is null/none/empty, the response channel
 is the channel the message came from.
 
-### ENV-10 — System prompt template substitution (coverage: test)
+### ENV-10 — Dynamic context reaches the system prompt (coverage: test)
 
-`message()` substitutes `{date}` (YYYY-MM-DD), `{time}`, `{memory}`
-(the assembled memory block — legacy memory string while the
-structured memory is inactive, see MEM-10) in the system prompt;
-`{news}` is replaced with the news file content when the configured
-file exists and stays literal when it does not.
+The system message carries the current date, time, news (when the
+configured file exists) and the memory block (legacy string while
+structured memory is inactive, see MEM-10). Since FDB-008 these live
+in a context suffix, not inline — see ENV-20; legacy `{date}`,
+`{time}`, `{news}`, `{memory}` placeholders in operator templates are
+stripped.
+
+### ENV-20 — Persona prefix is byte-stable (coverage: test)
+
+`message()` renders the system message as: static persona text
+(config template with all dynamic placeholders removed) followed by a
+`## Context` suffix holding date, time, news and memory. Two calls in
+the same channel produce byte-identical persona prefixes — the prompt
+cache can actually hit (the old inline `{date}`/`{time}` substitution
+invalidated it every minute).
 
 ### ENV-11 — Per-channel history shrink prefers busy channels (coverage: test)
 

specs/SPEC-003-safety.md

@@ -32,6 +32,12 @@ and game descriptions are attacker-influenced input.
 The `hack` envelope field remains as an advisory signal (logged,
 staff-notified) but is no longer the defense.
 
+### SAF-10 — Model calls carry a hashed user identifier (coverage: test)
+
+Chat calls pass `safety_identifier` = a short SHA-256 digest of the
+message author's name — OpenAI-side abuse tracing without shipping
+raw Discord identities (Codex review recommendation).
+
 ### SAF-04 — Hard daily budget, fail-closed (coverage: test)
 
 When `daily-budget-usd` is configured and today's estimated spend

specs/SPEC-006-operator-controls.md

@@ -57,6 +57,11 @@ the alert text is written to the error log — never silently dropped.
 loop; later: the FDB-011 scheduler); `!bot tasks on` restores.
 Bot-initiated posts also respect pause/quiet.
 
+### OPS-11 — Pins are listable (coverage: test)
+
+`!bot pins` answers with all pinned facts and their ids (global +
+per-channel) — without it, `!bot unpin <id>` required guessing ids.
+
 ### OPS-10 — Spend report (coverage: test)
 
 `!bot spend` answers in the staff channel with today's estimated

specs/SPEC-008-config.md

@@ -17,10 +17,11 @@ A responder bound to channel `X` uses `config["X"]` as its system
 prompt when that key exists, else `config["system"]`. One deployment
 can speak differently per channel.
 
-### CFG-03 — Missing news file leaves the placeholder untouched (coverage: test)
+### CFG-03 — Missing news file degrades silently (coverage: test)
 
-When `news` points to a non-existent file, the `{news}` placeholder
-stays literal in the system prompt (no crash, no empty substitution).
+When `news` points to a non-existent file, the context suffix simply
+carries no news section (no crash, no literal placeholder — revised
+with ENV-20; previously the `{news}` placeholder stayed literal).
 
 ### CFG-04 — Config hot-reload applies on the event loop (coverage: test)
 

specs/SPEC-010-behavior.md

@@ -0,0 +1,63 @@
+# SPEC-010 — Human-behavior layer
+
+The bot should feel like a considerate participant, not an instant
+wall of text: it decides *whether* to speak with a cheap classifier
+instead of trusting the main model's self-report, paces its replies,
+splits long answers, and sometimes just reacts. All knobs are
+per-deployment TOML; every feature degrades to the previous behavior
+when its knob is unset (config-off = v3.0.0 semantics).
+
+### BEH-01 — Classifier gates non-direct replies (coverage: test)
+
+With `classifier-model` configured, every non-direct user message
+first passes a cheap classification call (reply yes/no, factual
+yes/no, optional reaction emoji). `reply=false` means no main-model
+call happens at all — this is the boreness suppressor and the
+butting-into-conversations fix (replaces trusting `answer_needed`
+alone; the envelope flag still applies afterwards as second gate).
+
+### BEH-02 — Direct messages bypass the gate (coverage: test)
+
+Mentions and DMs never go through the classifier — someone addressing
+the bot always reaches the main model. Welcome and bot-initiated
+flows do not pass the gate either.
+
+### BEH-03 — Classifier failure fails open (coverage: test)
+
+A failed or unparseable classification (API error, budget refusal)
+falls through to the main model. Availability beats savings; the
+budget gate still protects spend.
+
+### BEH-04 — Reply pacing is typing-proportional (coverage: test)
+
+With `typing-chars-per-second` set (recommended 30), the typing
+indicator is held for `len(part) / cps` seconds per message part,
+capped at `typing-max-seconds` (default 8), before sending. Unset or
+0 = no pacing (v3.0.0 behavior).
+
+### BEH-05 — Factual answers skip the artificial delay (coverage: test)
+
+Messages the classifier tagged `factual` (opening hours, prices,
+addresses) are answered without the BEH-04 delay — utility beats
+theater exactly where users are waiting for information.
+
+### BEH-06 — Long answers are split (coverage: test)
+
+Answers longer than `split-threshold` chars (default 1200) are split
+at paragraph (then sentence) boundaries into at most
+`split-max-parts` (default 3) sequential messages, each under the
+Discord 2000-char limit (which unsplit answers would crash into
+today). Attached images go with the last part.
+
+### BEH-07 — Sometimes a reaction is the reply (coverage: test)
+
+When the classifier returns `reply=false` plus a reaction emoji, the
+bot adds that emoji to the user's message instead of staying fully
+silent. Zero main-model cost, human touch.
+
+### BEH-08 — Quiet hours stop bot-initiated posts (coverage: test)
+
+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.