Every commit scored for change-risk against this repo's own history, so 'elevated' means elevated here rather than on some global curve.
Needs review
43 commits sit in this repo's top risk tercile, which is 32% of the 133scored. The cut is drawn against this codebase's own history rather than a global curve, so a quiet repo still fills its top band, and here it starts at 8.1 out of 10. What pushes a commit up is size and spread together: a large change confined to one area scores below a smaller one scattered across a dozen files.
Commit categories over time, read off the subject line. Fixes carry the accent because that is the series this chart exists to show.
Began chore-led, now leaning feature.
Ranked by change-risk, highest first. Priority is a tercile of this repo's own distribution, so a quiet repo still fills its top band.
| # | Commit | Author | When | Lines | Risk | Top driver |
|---|---|---|---|---|---|---|
| 1 | d123bb17feat(ai-agents): fase 2 — camada de inteligência (intent classifier + 4 layers + RAG + memórias) | João Pedroclaudeassisted | 2mo ago | +5.6K -12 | 99%Elevated | more lines added than baseline |
| 2 | e68e3c0achore: initial project setup | João Pedro | 4mo ago | +15.4K -0 | 99%Elevated | more lines added than baseline |
| 3 | 852a2db5feat(ai-agents): client-ops tools da Sofia (ClickUp, n8n, Drive, agendamento) | tiago1002bravyclaudeassisted | 1mo ago | +1.7K -9 | 97%Elevated | more lines added than baseline |
| 4 | 656d2a37feat(automations): PR2-4 — engine completo (executor, actions, REST API) | João Pedroclaudeassisted | 2mo ago | +2.4K -29 | 97%Elevated | more lines added than baseline |
| 5 | 15cd047afeat: channel sync, webhook events, audio transcription and org preferences | João Pedro | 3mo ago | +4.3K -547 | 97%Elevated | more lines added than baseline |
| 6 | 27879bc8feat: enhance messaging and channel hub modules with new integrations | João Pedro | 4mo ago | +2.4K -27 | 97%Elevated | more lines added than baseline |
| 7 | b4f8bf22feat(ai-agents): MVP runtime — LLM service, tools, runner, router, CRUD | João Pedroclaudeassisted | 3mo ago | +1.9K -136 | 95%Elevated | more lines added than baseline |
| 8 | a3a00497feat(automations): PR1 — fundação do motor de automações (kill switch OFF) | João Pedroclaudeassisted | 2mo ago | +1.2K -75 | 94%Elevated | more lines added than baseline |
| 9 | be59c945feat: ratings, IAM, channel agents, api-keys refinements | João Pedro | 3mo ago | +1.2K -93 | 94%Elevated | more lines added than baseline |
| 10 | 05f62deffeat: api keys, public api, media resolver and dashboard/messaging refinements | João Pedroclaudeassisted | 3mo ago | +2.1K -541 | 94%Elevated | more lines added than baseline |
| 11 | 4cdfbd88fix(prompts): tool results gateados não confundem mais Sonnet 4.6 | João Pedroclaudeassisted | 2mo ago | +1.0K -5 | 92%Elevated | more lines added than baseline |
| 12 | 2f822811feat(pipelines): kanban com pipelines customizados a nível de org | João Pedroclaudeassisted | 3mo ago | +952 -0 | 92%Elevated | more lines added than baseline |
| 13 | 083c695brefactor(ai-agents): tool = provider, skill = LLM-callable function | João Pedroclaudeassisted | 3mo ago | +540 -574 | 90%Elevated | more lines added than baseline |
| 14 | f76bdd90feat(ai-catalog): CRUD endpoints for tools, skills, and agent attachments | João Pedroclaudeassisted | 3mo ago | +627 -1 | 90%Elevated | more lines added than baseline |
| 15 | 275afb42feat(ai-agents): skills + custom HTTP tools schema and runner integration | João Pedroclaudeassisted | 3mo ago | +661 -15 | 90%Elevated | more lines added than baseline |
| 16 | 249cbb88feat(ai-agents): fase 2.5 — wiring final da camada de inteligência | João Pedroclaudeassisted | 2mo ago | +571 -10 | 88%Elevated | more lines added than baseline |
| 17 | 903187f7feat(ai-agents/evals): plug runner no LLM real e adiciona CLI + CI | João Pedroclaudeassisted | 2mo ago | +582 -54 | 88%Elevated | more lines added than baseline |
| 18 | 39647d55feat(products): catálogo + skill getProductPitch + injeção no prompt | João Pedro | 3mo ago | +577 -1 | 88%Elevated | more lines added than baseline |
| 19 | 77ff5292feat(llm): migra OpenRouter → Anthropic SDK direto (corte de custo) | João Pedroclaudeassisted | 2mo ago | +500 -457 | 85%Elevated | more lines added than baseline |
| 20 | b61868c5feat(inbox-views): inbox views personalizadas por user (backend) | João Pedroclaudeassisted | 3mo ago | +459 -0 | 85%Elevated | more lines added than baseline |
| 21 | 8eb5b47bfeat(ai-agents): CUSTOM_SQL tool source — query Postgres direto | João Pedroclaudeassisted | 3mo ago | +592 -15 | 85%Elevated | more lines added than baseline |
| 22 | 02fa496afeat(ai-agents): schema foundation for AI agents | João Pedroclaudeassisted | 3mo ago | +399 -0 | 85%Elevated | more lines added than baseline |
| 23 | f6eedc14fix(http-tool-executor): normaliza emails (lowercase+trim) antes de chamar API | João Pedroclaudeassisted | 2mo ago | +442 -49 | 82%Elevated | more lines added than baseline |
| 24 | ac5bca36feat(ai-agents): jarvis stats endpoints | João Pedroclaudeassisted | 3mo ago | +364 -0 | 82%Elevated | more lines added than baseline |
| 25 | ebba0976feat(ai-agents): orchestrator tools — listAvailableAgents, delegateToAgent, handBackToOrchestrator | João Pedroclaudeassisted | 3mo ago | +336 -12 | 82%Elevated | more lines added than baseline |
| 26 | b6613056feat(ai-agents): organograma matricial ágil com hierarquia + departamento + squad | João Pedroclaudeassisted | 2mo ago | +271 -162 | 81%Elevated | more lines added than baseline |
| 27 | eae2dde2fix(ai-agents): cinto+suspensório contra "raciocínio verbalizado" | João Pedroclaudeassisted | 2mo ago | +215 -56 | 79%Elevated | more lines added than baseline |
| 28 | a7c39f3efeat(skills): gating de aprovação configurável por (agent, skill) | João Pedroclaudeassisted | 2mo ago | +256 -13 | 79%Elevated | more lines added than baseline |
| 29 | 6a332c94feat(messages): reply nativo cross-channel (Zappfy/Cloud API/IG fallback) | João Pedroclaudeassisted | 2mo ago | +207 -24 | 79%Elevated | more lines added than baseline |
| 30 | 1d2aa76bfeat(ai-agents): tool-failure observability + transient retry | João Pedroclaudeassisted | 2mo ago | +273 -40 | 79%Elevated | more lines added than baseline |
| 31 | 7161d7c2feat(inbox): filtros de não lidas/arquivadas + view default Archived | João Pedroclaudeassisted | 2mo ago | +221 -1 | 77%Elevated | more lines added than baseline |
| 32 | 0cfab8c4feat(ai-agents): brevidade radical + typing delay humanizado | João Pedroclaudeassisted | 3mo ago | +196 -10 | 77%Elevated | more lines added than baseline |
| 33 | 27ae3887feat(ai-agents): gate destructive tools via PendingAction (fase 2) | João Pedroclaudeassisted | 2mo ago | +179 -29 | 75%Elevated | more lines added than baseline |
| 34 | 6f4f5b7cfix(ai): handoff silencioso + erro real do LLM provider | João Pedroclaudeassisted | 2mo ago | +153 -83 | 75%Elevated | more lines added than baseline |
| 35 | 517129cffeat(zappfy): foto de perfil + auto-mark-as-read ao responder | João Pedro | 3mo ago | +160 -0 | 73%Elevated | more lines added than baseline |
| 36 | edfaf48dfeat(conversations): unread counter por user com realtime | João Pedro | 3mo ago | +173 -2 | 73%Elevated | more lines added than baseline |
| 37 | a335fe3bfeat(ai-agents): per-conversation AI toggle + org-wide AI settings | João Pedroclaudeassisted | 3mo ago | +149 -3 | 73%Elevated | more lines added than baseline |
| 38 | 609ddbadfeat(messages): endpoint DELETE pra deletar mensagem pra todos | João Pedroclaudeassisted | 2mo ago | +150 -0 | 71%Elevated | more lines added than baseline |
| 39 | 5996d8e2feat(media): padroniza resolução de mídia inbound entre canais | João Pedroclaudeassisted | 2mo ago | +156 -22 | 71%Elevated | more lines added than baseline |
| 40 | 93fa5682feat(catalog): migra catálogo de produtos pra Trivapp (single source of truth) | João Pedro | 2mo ago | +154 -49 | 71%Elevated | more lines added than baseline |
| 41 | 69b81be3feat(channels): visibility ORG/PRIVATE — privacidade real por canal | João Pedroclaudeassisted | 2mo ago | +191 -25 | 70%Elevated | more lines added than baseline |
| 42 | 91fb31cffeat(uploads): endpoint generico de upload de midia pra anexos do chat | tiago1002bravyclaudeassisted | 1mo ago | +98 -0 | 68%Elevated | more lines added than baseline |
| 43 | 26a5515ffeat(ai-agents): endpoint GET /ai-agents/watchdog/stats | João Pedroclaudeassisted | 2mo ago | +123 -0 | 68%Elevated | more lines added than baseline |
| 44 | c2580ba5fix(ai-agents): bloqueia URL inventada (whitelist por org) | João Pedroclaudeassisted | 2mo ago | +133 -3 | 66%Typical | more lines added than baseline |
| 45 | 1723fd51feat(ai-agents): built-in tool checkMembersAccess | João Pedroclaudeassisted | 2mo ago | +136 -0 | 66%Typical | more lines added than baseline |
| 46 | 846c25d7feat(ai): skill checkBonusEligibility — regra dos 7 dias do portal | João Pedroclaudeassisted | 2mo ago | +125 -0 | 66%Typical | more lines added than baseline |
| 47 | 74748818feat(conversations): endpoint pra pinar agent IA + engage atômico | João Pedro | 3mo ago | +108 -0 | 66%Typical | more lines added than baseline |
| 48 | c6fe9173fix(ai-agents): delegação atômica + listSkills retorna agents atribuídos | João Pedroclaudeassisted | 3mo ago | +121 -16 | 66%Typical | more lines added than baseline |
| 49 | cc03f921feat(inbox-views): filtros locais funcionam dentro de inbox views | João Pedroclaudeassisted | 2mo ago | +81 -7 | 63%Typical | more lines added than baseline |
| 50 | 2d26cc48fix(ai): aumenta debounce pra 8s e coalesce mensagens durante run | João Pedro | 2mo ago | +84 -41 | 63%Typical | more lines added than baseline |
Two views of the same model: where the cuts fall, and what commit shape lands you above them.
Every scored commit, binned on the raw 0 to 10 score rather than the percentile. Percentile ranks are uniform by construction, so that axis has no shape to draw. The dashed lines are the tercile cuts behind each row's priority pill.
The 133 most recent commits, on their own recency sample rather than the feed above: that defaults to risk-sorted, so reusing it would plot only the top tercile and call it the spread. Big and scattered is what the model penalises. Click a dot to open it.
kursku/chat-bullq-api has 144 commits in its history from 2 contributors, the first of them Mar 27, 2026. In the last 90 days 166 files were touched, 316 times in total, most often prompt-builder.service.ts. Every commit is scored for change risk from its size, spread and the history of the files it touches.
