v0.1.8

What's Changed

• feat(memory): add session_search MCP tool (Layer-3 memory) by @ZhePang in #245
• fix(web): parse session_search call-signature args, not just JSON by @ZhePang in #246
• refactor(memory): write-side guidance around session_search routing by @ZhePang in #247
• feat(chat): render chained chat turns as one conversation by @ZhePang in #250
• feat(memory): /memory/eval dashboard — Layer A telemetry surface by @ZhePang in #249
• feat(chat): stream working steps on top, collapse into a dropdown on completion by @ZhePang in #251
• feat(memory-eval): calendar popover for "compare around"; default to today by @ZhePang in #252
• feat(chat): spotlight the newest step in the live working box by @ZhePang in #253
• fix(runtime-ws): daemon liveness watchdog + replay pending on reconnect by @ZhePang in #254

Full Changelog: v0.1.7...v0.1.8

v0.1.7

What's Changed

• chore: extract Beevibe Crystal to its own repo by @ws2694 in #238
• feat(session): conversation_id + active-sessions filter excludes chat by @ZhePang in #239
• feat(watch-tasks): team-agent wake-up on dispatched task completion by @ZhePang in #240
• fix(watch-tasks): inherit runtime_id on the wake + hide system wakes from user-bubble rendering by @ZhePang in #241
• feat(chat): render watch-fired event as a system pill in chat history by @ZhePang in #242
• refactor(watch-pill): style as an event row, not a bordered bubble by @ZhePang in #243

Full Changelog: v0.1.6...v0.1.7

v0.1.6

What's Changed

• feat(chat): inline live IC transcripts under create_task by @ws2694 in #200
• docs(tools): sharpen update_core_memory description (append-vs-replace + dated-update ban) by @ZhePang in #204
• fix(dashboard): give 'cancelled' tasks their own legend bucket by @ZhePang in #206
• daemon: --config-root flag for dev-only multi-instance by @ZhePang in #208
• daemon: add ISO 8601 timestamps to logger by @ZhePang in #219
• fix(team-agent): respond-variant lifecycle + verify-before-acting on delegated work by @ZhePang in #220
• fix(mcp-config): 10-min per-tool-call timeout across Claude Code + codex by @ZhePang in #221
• feat(chat): persist intermediate steps in conversation history by @ZhePang in #222
• Revert "feat(chat): persist intermediate steps in conversation history" by @ZhePang in #223
• fix(chat): seed "latest" cache slot before stripping ?new=1 to stop mid-session blank by @ZhePang in #224
• feat(escalation): human review page with proposal backfill by @ZhePang in #225
• feat(mesh): negotiation detail page + per-kind row routing by @ZhePang in #226
• feat(mesh): time-window filter (24h / 7d / 30d / All) by @ZhePang in #227
• feat(crystal): import Beevibe Crystal package by @ws2694 in #228
• docs: clarify Beevibe Crystal setup by @ws2694 in #229
• fix(mesh): qualify window WHERE fragments with table aliases by @ZhePang in #232
• fix(chat): keep streaming surface up after 504 chat_turn_timeout by @ZhePang in #233

Full Changelog: v0.1.5...v0.1.6

v0.1.5

What's Changed

• fix(skills): register beevibe-verify-pr in tier filter by @ZhePang in #194
• fix(dashboard): active_sessions KPI counts running sessions, not active agents by @ZhePang in #195
• chore: type SessionRow.deleted_at + drop runtime/register success log by @ZhePang in #183
• fix(claim): honor task priority when claiming pending sessions by @ZhePang in #196

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• chore(routing): drop team-agent disallowed-tools plumbing by @ZhePang in #159
• chore(chat): remove per-turn runtime picker from the chat composer by @ZhePang in #162
• feat(chat): surface runtime-pinning to the user by @ZhePang in #163
• feat(web): surface beevibe-daemon sync in the runtimes panel + agent detail by @ZhePang in #165
• feat(agents): list view as default, with inline runtime/model/review pickers by @ws2694 in #166
• feat(agents): chip-style pickers for Runtime / Model / Review policy by @ws2694 in #167
• style(web): frosted-chrome UI polish (sidebar, chat, agents) by @ws2694 in #168
• fix(chat): first click on "New chat" was a no-op when FRESH slot was stale by @ws2694 in #169
• style(chat): swap honeycomb + dot-dot-dot for spinning icosahedron by @ws2694 in #170
• style(web): translucent surfaces to match the glassy buttons by @ws2694 in #171
• style(web): refresh agent cards with DiceBear avatars by @ws2694 in #173
• style(web): add DiceBear thumbs user avatars by @ws2694 in #174
• fix(prompts): three-lane team-agent routing + universal injection across chat/task/MCP by @ZhePang in #175
• docs: add Star History chart to README by @ZhePang in #176
• style(web): simplify chat history sidebar by @ws2694 in #177
• style(web): polish chat working state by @ws2694 in #180
• fix(routing): tell team agents to check task state before reporting on it by @ZhePang in #179
• style(web): make new chat button honey yellow by @ws2694 in #178
• fix(skills): pre-task-setup prunes landed worktrees before adding a new one by @ZhePang in #182
• feat: capability network MVP by @ws2694 in #153
• fix(claim): enforce per-agent task cap + fix task-status lying during cap-block by @ZhePang in #186
• fix(deploy): rename stale pgmigrations row before migrate up by @ws2694 in #187
• fix(deploy): run migrations from api startCommand, ship node-pg-migrate in image by @ws2694 in #188
• fix(deploy): pass --no-check-order so out-of-order capability migration applies by @ws2694 in #189
• fix(task): auto-recover stuck task sessions + add manual Retry button (#129) by @ZhePang in #190
• feat(skills): beevibe-verify-pr — block update_progress(done) on red CI by @ZhePang in #191
• feat(task): retry from cancelled + Cancel/Retry in side panel by @ZhePang in #192
• fix(release): pre-build @beevibe/sandbox before bun-bundle daemon by @ZhePang in #193

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• fix(mesh): reject caller's ask/negotiate fast when callee session fails by @ZhePang in #135
• fix(tasks): scope list to caller's owner tree, always show archive, autofit lanes by @ZhePang in #136
• fix(cancel): route signal to daemon, block late update_progress, drop misleading session button by @ZhePang in #137
• fix(memory): stable scope-tab counts via dedicated counts endpoint by @ZhePang in #138
• fix(mesh): fast-fail across processes via session.updated pg_notify by @ZhePang in #140
• fix(tasks): suppress 'No tasks yet' during background refetch by @ZhePang in #142
• fix(inbox): scope task branches by owner, not direct-person-creator by @ZhePang in #143
• fix(task): human revise dispatches the revision session by @ZhePang in #144
• fix(inbox): invalidate inbox cache on task approve/reject/revise/cancel by @ZhePang in #145
• ui: natural chat UI + knowledge-graph agent view by @ws2694 in #146
• fix(chat): wire client session_id so live steps appear while agent is running by @ws2694 in #156
• fix(routing): team agents cannot handle specialist work directly by @ws2694 in #154
• feat(runtime): recover pluggable runtimes and add codex by @ws2694 in #150
• fix(ui): strip task XML wrapper from session H1, wrap transcript content by @ws2694 in #158
• feat: persist work product body + daemon fetch resilience by @ws2694 in #157
• fix(daemon): claimer test passes runtimeRegistry to satisfy ClaimerConfig by @ZhePang in #160
• feat(daemon): sync subcommand for registering newly-installed CLIs by @ZhePang in #161

Full Changelog: v0.1.2...v0.1.3

v0.1.2

What's Changed

• deploy: release workflow auto-updates release branch on stable tags by @ZhePang in #86
• deploy: point Railway badge at published template URL by @ZhePang in #87
• deploy: README — document Railway template post-deploy config step by @ZhePang in #89
• Landing site rewrite — team-first positioning, bee mascot assets, Apache-2.0 framing by @ws2694 in #85
• Remove infra/landing/ — marketing landing moved to a private repo by @ws2694 in #93
• fix(agents): split runtime/model fields in AgentDisplay by @ZhePang in #96
• fix(agents): drop model="opus" from DEFAULT_RUNTIME_CONFIG by @ZhePang in #97
• fix(agents): replay PR #98 onto main (model edit affordance) by @ZhePang in #100
• feat(memory): per-block descriptions + tag_line + persistent-specialist framing by @ZhePang in #99
• fix(agents): split identity (system_prompt_addition) from description (persona block) by @ZhePang in #101
• feat(agents): per-agent review policy edit + default to auto_done by @ZhePang in #102
• fix(agents): sync runtime_config.type with bound runtime's CLI by @ZhePang in #103
• frontend fixes — clickable agent sessions + restore api typecheck by @ws2694 in #94
• fix(migration): rename usage-window-index to resolve timestamp collision by @ws2694 in #105
• chore(scripts): add db:reset for one-command local DB reset by @ws2694 in #106
• fix(web): make agent detail read-only when caller doesn't own the agent by @ZhePang in #104
• Reposition README around team OS narrative by @ws2694 in #107
• Add OpenCode runtime by @ws2694 in #108
• poc: sandboxed repo-powered agent tools by @ws2694 in #95
• Revert "poc: sandboxed repo-powered agent tools" by @ZhePang in #110
• Revert "Add OpenCode runtime" by @ZhePang in #111
• feat(agents): show owner on agent detail page and side panel by @ZhePang in #113
• feat(web): owner can edit agent core memory blocks via UI by @ZhePang in #112
• fix(agent-network): show every owner's team, not just shared-room peers by @ZhePang in #115
• fix(runtimes): share install instructions and refresh presence on WS connect by @ZhePang in #114
• fix(chat): keep user message visible on 503; auto-recover when daemon reconnects by @ZhePang in #116
• fix(hub): isOnline considers heartbeat freshness, not just WS push by @ZhePang in #117
• fix(panel): make agent peek-panel recent sessions clickable by @ZhePang in #118
• fix(memory): per-fact-type descriptions surface failure-mode guards (issue #90 part 1) by @ZhePang in #119
• feat(memory): issue #90 fixes B + D — re-read work products, owner delete by @ZhePang in #120
• fix(memory): fact scope reflects saver's hierarchy_level, not hardcoded 'ic' by @ZhePang in #121
• fix(agents): specialization comes from tag_line only, no domain fallback by @ZhePang in #122
• fix(chat): thinking indicator survives navigation away + back by @ZhePang in #123
• chore(scripts): prune milestone e2e scripts, fix stale executor prose by @ZhePang in #124
• docs: refresh package READMEs to match daemon-first architecture by @ZhePang in #125
• docs: restructure root README around positioning + one-command quickstart by @ZhePang in #126
• feat(agents): show CLI-to-MCP setup on the team agent detail page by @ZhePang in #132
• fix(chat): thinking indicator survives nav-away and clears after response by @ZhePang in #131
• fix(daemon): WORKSPACE_ROOT footgun + diagnostics + UI failure mapping by @ZhePang in #134

Full Changelog: v0.1.1...v0.1.2

v0.1.1

What's Changed

• Phase 7 PR C — welcome wizard install UX (brew / npx / direct download) by @ZhePang in #83

Full Changelog: v0.1.0...v0.1.1

v0.1.0

What's Changed

• M1 — core data layer: domain, ports, Postgres adapter by @ZhePang in #1
• ci: GitHub Actions workflow for build, typecheck, lint, test by @ZhePang in #11
• M2 — Claude Code runtime + workspace provisioner by @ZhePang in #10
• M3 — services layer + pgvector + LLM adapters + agent-driven memory by @ZhePang in #12
• M4 — core auth (bv_a_/bv_u_ API keys + caller resolution) by @ZhePang in #13
• M5 — executor binary (polling, dispatch, claim, reap, review-policy gate) by @ZhePang in #14
• M8 — design mocks for the web frontend by @ws2694 in #15
• M8 — packages/web (scaffold + shell + tasks list) by @ws2694 in #18
• design(web): mock-fidelity pass + mesh sidebar entry by @ws2694 in #24
• design(web): mock-fidelity pass, notion sidebar/kanban, drop fixtures by @ws2694 in #26
• M8 — web API contract scaffold + detail pages by @ws2694 in #27
• M6 — API server (MCP tools + REST + MeshServer + escalation) + prod-style dev orchestrator by @ZhePang in #28
• M8 — wire mutations to M6 + add read-views layer (#30) by @ws2694 in #31
• M8.A + M8.E — POST /task + remove aspirational threads (re-PR against main) by @ws2694 in #38
• M8.B — GET /dashboard with data/display split (#33) by @ws2694 in #39
• M8.C + M8.D — GET /mesh phase 1 + GET /promotion + audit log (closes #34, #35) by @ws2694 in #42
• M8 final integration: SSE, + New task modal, briefing persistence, agent metrics (#45) by @ws2694 in #48
• M8 cleanup — remove no-op + New agent affordances by @ws2694 in #49
• M7 — full-stack integration test + manual smoke recipe by @ZhePang in #51
• feat: M9 — agent skill system + memory tooling refresh (closes #9) by @ZhePang in #54
• docs: per-package READMEs + open-source root README by @ZhePang in #59
• Phase 1: schema scaffold + session_event + cli normalize by @ZhePang in #72
• Phase 2: per-user SSE isolation + dev orchestration by @ZhePang in #73
• Phase 3: DispatchService + Daemon/Runtime repos (additive infra) by @ZhePang in #74
• Phase 4: daemon-first runtime — endpoints + binary + chat by @ZhePang in #75
• Phase 5 — daemon-first frontend + m8 UI parity + password auth by @ZhePang in #77
• License: Apache 2.0 + trademark policy + CLA/DCO contributor flow by @ZhePang in #78
• Phase 7 PR A — Railway one-click + Docker self-host foundation by @ZhePang in #79
• Phase 7 PR B — daemon distribution (native binaries, brew tap, npm publish, auto-update) by @ZhePang in #80
• deploy: switch npm publish to Trusted Publishing (OIDC) by @ZhePang in #81
• deploy: fix release workflow — npm 11.5+ for Trusted Publishing, pre-release detection by @ZhePang in #82

New Contributors

• @ZhePang made their first contribution in #1

Full Changelog: https://github.com/beevibe-ai/beevibe/commits/v0.1.0

v0.0.1-rc2

What's Changed

• deploy: fix release workflow — npm 11.5+ for Trusted Publishing, pre-release detection by @ZhePang in #82

Full Changelog: v0.0.1-rc1...v0.0.1-rc2