Chat Phase 1 — multi-agent routing, mentions, policy, reactions, bare-slash guardrail by jaylfc · Pull Request #234 · jaylfc/tinyagentos

feat(chat): channel settings defaults + mutation helpers for Phase 1 routing

jaylfc

committed

feat(chat): mention parser (@slug/@all/@humans) for router fanout

jaylfc

committed

feat(chat): GroupPolicy — per-agent cooldown + per-channel rate cap

jaylfc

committed

feat(chat): context-window builder for per-bridge rolling history

jaylfc

committed

feat(chat): normalise message.metadata to dict; support hops_since_user field

jaylfc

committed

feat(chat): router fanout — mentions + quiet/lively modes + hop cap + policy gates

jaylfc

committed

fix(chat): log context-fetch failure in router instead of silent swallow

jaylfc

committed

feat(chat): wire GroupPolicy onto app.state for router

jaylfc

committed

feat(chat): slash-command registry with 11 commands (mute/unmute/leave/summon/quiet/lively/hops/cooldown/topic/rename/help)

jaylfc

committed

feat(chat): intercept slash commands in POST /api/chat/messages

jaylfc

committed

feat(chat): semantic reactions — 👎 regenerate, 🙋 wants-reply registry

jaylfc

committed

feat(chat): reactions POST/DELETE + wants_reply GET endpoints

jaylfc

committed

feat(chat): event payload carries hops/force_respond/context; replies re-dispatch for agent-to-agent fanout

jaylfc

committed

feat(bridges): honor force_respond, render context, suppress NO_RESPONSE

jaylfc

committed

docs: correct beta date — week of Mon 20 April (not 21)

jaylfc

committed

ci: mark rolling-images as pre-release (it's a build artifact, not a user release)

jaylfc

committed

test(chat): integration coverage for @all, cooldown, cooldown-override-by-mention, DM force-respond

jaylfc

committed

test(e2e): agents must reference each other in lively roundtable

jaylfc

committed

ci: publish openclaw base images to jaylfc/tinyagentos-images, not main repo

jaylfc

committed

Merge branch 'master' into feat/chat-phase-1-conversational-core

jaylfc

committed

refactor(chat): drop slash-command intercept — / passes through to agents

jaylfc

committed

feat(chat): reject bare slash in non-DM channels (must @<agent> or @all)

jaylfc

committed

feat(chat): REST admin endpoints for channel settings/members/muted

jaylfc

committed

docs: spec reflects UI-driven channel admin (drop slash intercept)

jaylfc

committed

fix(bridges): thread channel_id through post_reply in langroid/pocketflow/openai-agents

jaylfc

committed

fix(chat): major review fixes (atomic policy, regen context, e2e test)

jaylfc

committed

style: minor review cleanups (E702, token clamp, spec corrections)

jaylfc

committed

docs: scrub remaining slash-command references from spec

jaylfc

committed

Merge branch 'master' into feat/chat-phase-1-conversational-core

jaylfc

committed

fix(chat): address final 3 CodeRabbit findings (hops coerce, context hoist, regenerate prompt)

jaylfc

committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Chat Phase 1 — multi-agent routing, mentions, policy, reactions, bare-slash guardrail#234

Chat Phase 1 — multi-agent routing, mentions, policy, reactions, bare-slash guardrail#234
jaylfc merged 30 commits intomasterfrom
feat/chat-phase-1-conversational-core

Commits on Apr 19, 2026

feat(chat): channel settings defaults + mutation helpers for Phase 1 routing

feat(chat): mention parser (@slug/@all/@humans) for router fanout

feat(chat): GroupPolicy — per-agent cooldown + per-channel rate cap

feat(chat): context-window builder for per-bridge rolling history

feat(chat): normalise message.metadata to dict; support hops_since_user field

feat(chat): router fanout — mentions + quiet/lively modes + hop cap + policy gates

fix(chat): log context-fetch failure in router instead of silent swallow

feat(chat): wire GroupPolicy onto app.state for router

feat(chat): slash-command registry with 11 commands (mute/unmute/leave/summon/quiet/lively/hops/cooldown/topic/rename/help)

feat(chat): intercept slash commands in POST /api/chat/messages

feat(chat): semantic reactions — 👎 regenerate, 🙋 wants-reply registry

feat(chat): reactions POST/DELETE + wants_reply GET endpoints

feat(chat): event payload carries hops/force_respond/context; replies re-dispatch for agent-to-agent fanout

feat(bridges): honor force_respond, render context, suppress NO_RESPONSE

docs: correct beta date — week of Mon 20 April (not 21)

ci: mark rolling-images as pre-release (it's a build artifact, not a user release)

test(chat): integration coverage for @all, cooldown, cooldown-override-by-mention, DM force-respond

test(e2e): agents must reference each other in lively roundtable

ci: publish openclaw base images to jaylfc/tinyagentos-images, not main repo

Merge branch 'master' into feat/chat-phase-1-conversational-core

refactor(chat): drop slash-command intercept — / passes through to agents

feat(chat): reject bare slash in non-DM channels (must @<agent> or @all)

feat(chat): REST admin endpoints for channel settings/members/muted

docs: spec reflects UI-driven channel admin (drop slash intercept)

fix(bridges): thread channel_id through post_reply in langroid/pocketflow/openai-agents

fix(chat): major review fixes (atomic policy, regen context, e2e test)

style: minor review cleanups (E702, token clamp, spec corrections)

docs: scrub remaining slash-command references from spec

Merge branch 'master' into feat/chat-phase-1-conversational-core

fix(chat): address final 3 CodeRabbit findings (hops coerce, context hoist, regenerate prompt)