implement
Skill0 starsChanged 27 days agoMITView on GitHub
Etapa 4 do Fluxo Smart — executa as tasks via workflow smart-implement com escada de modelos (haiku→sonnet→opus→fable) e gates G0/G1 por task
191 other projects use a version of this.
Adopted by
Most recently taken up by
- revisit-studies/studytheir own version, first seen Nov 9, 2026
- dz114879/skillstheir own version, first seen Nov 8, 2026
- Dimon94/dveritytheir own version, first seen Nov 8, 2026
- Dimon94/skillstheir own version, first seen Nov 8, 2026
- The-Vibe-Company/quivrtheir own version, first seen Nov 8, 2026
- miklschmidt/archboardtheir own version, first seen Nov 8, 2026
Used with
Projects with it also use setup-ts-deep-modules, ask-matt and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
liufeng1976Skillsetup-ts-deep-modules
Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.
Also in 43 projects that use this
GoogleSkillask-matt
Ask which skill or flow fits your situation. A router over the skills in this repo.
Also in 91 projects that use this
RoentekSkillloop-me
Grill me about specs for the workflows I want to build, within this workspace.
Also in 50 projects that use this
AWSSkillto-questionnaire
Turn a decision you can't fully answer into a questionnaire for someone else to fill in.
Also in 88 projects that use this
ryotaroh180105Skillclaude-handoff
Hand the current conversation off to a fresh background agent that picks up the work immediately.
Also in 42 projects that use this
GoogleSkillto-spec
Turn the current conversation into a spec and publish it to the project issue tracker — no interview, just synthesis of what you've already discussed.
Also in 168 projects that use this
- +3 more from google
Other versions191 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 191
- mattpocock/skills★ 309ksame name
- QuivrHQ/quivr★ 40ksame name
- The-Vibe-Company/quivr★ 40ksame name
- foambubble/foam★ 17ksame name
- formbricks/formbricks★ 13ksame name
- ArcReel/ArcReel★ 6.3ksame name
- vinvcn/mattpocock-skills-zh-CN★ 5.5ksame name
- fallow-rs/fallow★ 5.3ksame name
- withcoral/coral★ 4.9ksame name
- bestofjs/bestofjs★ 3.1ksame name
181 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository28 other files it gives its agents.
- ado-closeSkill
- ado-commentSkill
- ado-getSkill
- ado-refineSkill
- adrSkill
- approve-reviewSkill
- appSkill
- autoSkill
- challenge-specSkill
- completeSkill
- dart-clean-archSkill
- discoverySkill
- fixSkill
- flutter-clean-archSkill
- jira-refineSkill
- kickoffSkill
- linear-refineSkill
- orcamentoSkill
- planSkill
- post-reviewSkill
- pr-openSkill
- pr-statusSkill
- refineSkill
- reviewSkill
- specifySkill
- statusSkill
- tasksSkill
- verifySkill
The file
Files
Esta skill não implementa: ela prepara os args, dispara o workflow smart-implement e trata o resultado. Quem escreve código são os dev-implementer no tier que a escada decidir.
1. Detectar projeto
PROJECT_PATH=$(git rev-parse --show-toplevel 2>/dev/null || pwd)
source ~/.claude/bin/get-project.sh
PROJECT_NAME=$(get-project-name)
WF_DIR="$HOME/.claude/workflow/$PROJECT_NAME"2. Pré-requisitos
Leia $WF_DIR/plan.md, $WF_DIR/tasks.md e $WF_DIR/current.json. Faltando algum, informe a etapa.
Valide o plano de tasks antes de disparar (campos da escada, Red-only, testes que nenhuma task cria):
python3 ~/.claude/bin/wf_gate.py tasks-check --workflow-dir "$WF_DIR" --repo "$PROJECT_PATH"Saída 1: mostre os erros e pare, apontando /tasks — sem isso o G0 reprova por construção.
3. Stack
Escolha o profile em ~/.claude/stacks/*.json cujo detect existe na raiz do repo ou no package root mais próximo dos affects (Flutter: pubspec.yaml; app em subdiretório como app/ conta). Monorepo com mais de uma stack: use a do diretório onde a maioria dos affects cai. Nenhum profile casa → pare e diga qual stack falta (o motor não roda sem G0).
3a. Audit do harness
python3 ~/.claude/bin/harness-audit.py --repo "$PROJECT_PATH" --stack <stack> --phase kickoff --textRoda os comandos do G0 num arquivo de controle e confere SDK, deps, codegen, submódulos e o profile contra o repo.
- FAIL (critical) → pare. Mostre os findings e ofereça
--fix(link do SDK,make setup/pub get + build_runner, submódulo limpo). Só rode--fixcom OK do usuário; depois, audite de novo. - major → mostre numa linha cada e siga. Submódulo que guarda o
adr_dirfora do commit registrado merece destaque: os ADRs lidos não são os desta branch. - Falha de ambiente que escapar do audit e aparecer no G0 depois vira check novo no
harness-audit.py, não lesson.
4. Preparar run
bash ~/.claude/bin/capture-metrics.sh start implement "$PROJECT_NAME" "$PROJECT_PATH"
RUN_ID="impl-$(date -u +%Y%m%dT%H%M%SZ)"
RUN_DIR="$WF_DIR/runs/$RUN_ID"
mkdir -p "$RUN_DIR"Relatório: início da etapa
python3 ~/.claude/bin/wf-report.py stage-start implement --workflow-dir "$WF_DIR" ${CLAUDE_FLOW_SESSION_ID:+--session "$CLAUDE_FLOW_SESSION_ID"} || trueCheckpoint base:
current.json.exec.checkpointexiste (retomada ou execução anterior) → use-o.- Senão:
bash ~/.claude/bin/wf-checkpoint.sh create "$PROJECT_PATH"e grave emexec.checkpointeexec.base_checkpoint(o verify usa o base para o diff do ciclo inteiro).
Grave o profile de stack que vai nos args (com overrides do run, ex.: autofix: [] em repo não formatado) em $RUN_DIR/stack.json: o G0 lê esse arquivo via --profile e só cai no ~/.claude/stacks/<stack>.json se ele não existir.
Atualize status para "implementing".
5. Disparar o workflow
Chame a ferramenta Workflow com name: "smart-implement" e args (objeto JSON, não string):
{
"project_path": "<PROJECT_PATH>",
"workflow_dir": "<WF_DIR>",
"run_dir": "<RUN_DIR>",
"stack_name": "flutter",
"stack": { "...conteúdo de ~/.claude/stacks/flutter.json..." },
"checkpoint": "<exec.checkpoint>",
"plan_path": "<WF_DIR>/plan.md",
"spec_path": "<WF_DIR>/spec.md",
"lessons_path": "<~/.claude/projects/PROJECT_NAME/lessons.md, se existir>",
"tasks": [ "...items com status != done, na ordem do tasks.md, com id/title/description/complexity/risk/tier0/tier/attempts/tests/affects..." ],
"max_attempts": 5,
"per_tier": 2
}O workflow roda em background. Informe em 1 linha: "Rodando smart-implement (N tasks, run <RUN_ID>). Acompanhe em /workflows." e encerre o turno.
6. Ao receber o resultado
Grave o JSON retornado em $RUN_DIR/workflow-result.json e persista:
python3 ~/.claude/bin/wf-event.py persist --run-dir "$RUN_DIR" --workflow-dir "$WF_DIR" --result-file "$RUN_DIR/workflow-result.json"
python3 ~/.claude/bin/wf-report.py import-run implement --workflow-dir "$WF_DIR" --run-dir "$RUN_DIR" || true
bash ~/.claude/bin/capture-metrics.sh end implement "$PROJECT_NAME" "$PROJECT_PATH"Marque [x] no tasks.md para as tasks done. Apresente:
## Implement — <status>
| Task | Cx | tier0 → final | Tentativas | Status |
|---|---|---|---|---|
- Escaladas: N · tokens de saída (trace): XRelatório: fim da etapa. Escreva com Write um resumo de 3 a 10 linhas em $WF_DIR/.stage-summary.md (nunca interpole texto em shell) e rode (acrescente --status blocked em backtrack/blocked):
python3 ~/.claude/bin/wf-report.py stage-end implement --workflow-dir "$WF_DIR" --summary-file "$WF_DIR/.stage-summary.md" || trueBacktracks e blockers já entram pelo import-run; use decision só para escolhas suas fora disso.
Depois, por status:
- done →
status: "implemented". Auto on: siga para/verify. Auto off: "Implementado. Verificar?/verify". - backtrack → o dev-implementer declarou o plano inviável na task
blocked_task. Aplique o protocolo de desvio: mostre oreason, proponha o ajuste noplan.md(espec.mdse for o caso) e faça umAskUserQuestion(parada incondicional, mesmo com auto on) com "Aplicar o ajuste (Recommended)", "Ajustar" e "Rejeitar". Aprovado, registre embacktrackse rode/implementde novo (retoma das pendentes). - blocked → escada esgotada, falha repetida entre tiers ou gate quebrado. Mostre o
reasone as hipóteses dodiagnosisordenadas por confidence (lente, hipótese, evidência, ação recomendada). Parada incondicional: faça umAskUserQuestioncom as opções corrigir spec →/challenge-spec; replanejar →/plan; corrigir ambiente e retomar →/implement; assumir a task manualmente. Não decida sozinho, nem com auto on.
Não fazer
- Não implemente tasks no contexto principal "para adiantar" — isso fura a escada e a telemetria.
- Não commite. O checkpoint é tree object, não commit.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

