factory-metrics
Skill0 starsChanged 6 months agoMITView on GitHub
Routine Tier 1 que coleta metricas da fabrica de software: tempo spec->PR, PRs/dia, taxa de correcao autonoma, taxa de bloqueador humano. Armazena em .claude/factory-metrics.json e exibe relatorio semanal. Invocar uma vez -- o loop se auto-agenda via ScheduleWakeup.
In this repository27 other files it gives its agents.
- frontend/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- frontend/CLAUDE.mdCLAUDE.md
- add-entity-to-auditSkill
- auditSkill
- babysit-prsSkill
- batchSkill
- daily-summarySkill
- feature-frontSkill
- featureSkill
- init-projectSkill
- migrateSkill
- planSkill
- register-screenSkill
- review-archSkill
- review-frontSkill
- review-prSkill
- setup-architectureSkill
- setup-designSkill
- setup-infraSkill
- shipSkill
- tasksSkill
- watch-ciSkill
- write-jobSkill
- write-migrationSkill
- write-reportSkill
- write-testSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 10
No inicio de cada iteracao, capturar contexto usando o Bash tool com o comando pwd para obter $repoRoot.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Voce e o factory-metrics do projeto financas-lab. Verifique se e hora de coletar metricas da fabrica e agende a proxima verificacao.
No inicio de cada iteracao, capturar contexto usando o Bash tool com o comando
pwd para obter $repoRoot. Definir tambem:
$stateFile = "$repoRoot/.claude/factory-metrics.json" e $agora = Get-Date
Passo 0 -- Capturar contexto
$repoRoot = (pwd).Path
$stateFile = "$repoRoot/.claude/factory-metrics.json"
$agora = Get-DatePasso 1 -- Verificar se e hora de coletar
$state = if (Test-Path $stateFile) {
Get-Content $stateFile | ConvertFrom-Json
} else {
[PSCustomObject]@{ ultimaExecucao = [datetime]::MinValue.ToString('o'); prs = @() }
}
$ultimaExecucao = [datetime]$state.ultimaExecucao
$horasDesdeUltima = ($agora - $ultimaExecucao).TotalHoursSe $horasDesdeUltima -lt 20: reportar "factory-metrics: proximo em
$([math]::Round(20 - $horasDesdeUltima, 1))h" e ir direto para Passo 5
(agendar proxima iteracao). Nao coletar metricas.
Iteracao (apenas se >= 20h desde ultima execucao)
Passo 2 -- Coletar PRs mergeados nos ultimos 7 dias ainda nao no state
$dataLimite = $agora.AddDays(-7).ToString('yyyy-MM-ddTHH:mm:ssZ')
$prsMergeados = gh pr list --state merged --limit 50 `
--json number,title,headRefName,createdAt,mergedAt `
| ConvertFrom-Json `
| Where-Object {
[datetime]$_.mergedAt -gt [datetime]$dataLimite -and
($_.headRefName -match '^(feat|fix|docs)/')
}
# Filtrar apenas PRs ainda nao no state
$prsRegistrados = @($state.prs | ForEach-Object { $_.pr_number })
$prsNovos = @($prsMergeados | Where-Object { $_.number -notin $prsRegistrados })Passo 3 -- Calcular metricas para cada PR novo
foreach ($pr in $prsNovos) {
# Obter commits do PR
$commitsData = gh pr view $pr.number --json commits | ConvertFrom-Json
$todosCommits = @($commitsData.commits)
$primeiroCommit = $todosCommits | Sort-Object authoredDate | Select-Object -First 1
# tempo_spec_pr_min: diferenca entre abertura do PR e primeiro commit (proxy de "work to PR").
# Negativo = clock skew ou PR aberto antes do primeiro commit -> registrado como null.
$tempoMin = if ($primeiroCommit) {
$diff = ([datetime]$pr.createdAt - [datetime]$primeiroCommit.authoredDate).TotalMinutes
if ($diff -lt 0) { $null } else { [math]::Round($diff, 0) }
} else { $null }
$commitsFix = @($todosCommits | Where-Object {
$_.messageHeadline -match '^fix\('
}).Count
$prBody = gh pr view $pr.number --json body | ConvertFrom-Json
$teveBloqueador = [bool]($prBody.body -match 'BLOQUEADOR')
# teve_correcao_autonoma: heuristica -- commits fix( presentes E mais de 1 commit total.
# Detecta "iteracao com correcao" mas nao distingue correcao por feedback humano vs autonoma.
$entrada = [PSCustomObject]@{
pr_number = $pr.number
titulo = $pr.title
branch = $pr.headRefName
aberto_em = $pr.createdAt
mergeado_em = $pr.mergedAt
tempo_spec_pr_min = $tempoMin
commits_total = $todosCommits.Count
commits_fix = $commitsFix
teve_correcao_autonoma = ($commitsFix -gt 0 -and $todosCommits.Count -gt 1)
teve_bloqueador = $teveBloqueador
}
$state.prs += $entrada
}Passo 4 -- Exibir relatorio
Calcular metricas a partir de $state.prs filtrado para os ultimos 7 dias:
$prs7d = @($state.prs | Where-Object {
[datetime]$_.mergeado_em -gt $agora.AddDays(-7)
})
$tempos = @($prs7d | Where-Object { $_.tempo_spec_pr_min -ne $null } |
ForEach-Object { $_.tempo_spec_pr_min })
$mediaTempo = if ($tempos.Count -gt 0) {
[math]::Round(($tempos | Measure-Object -Average).Average, 0)
} else { 'N/A' }
$medianaTempo = if ($tempos.Count -gt 0) {
$sorted = $tempos | Sort-Object
$mid = [math]::Floor($sorted.Count / 2)
if ($sorted.Count % 2 -eq 0) {
[math]::Round(($sorted[$mid-1] + $sorted[$mid]) / 2, 0)
} else { $sorted[$mid] }
} else { 'N/A' }
$correcaoCount = @($prs7d | Where-Object { $_.teve_correcao_autonoma }).Count
$bloqueadorCount = @($prs7d | Where-Object { $_.teve_bloqueador }).Count
$totalPrs7d = $prs7d.Count
$prsDia = if ($totalPrs7d -gt 0) { [math]::Round($totalPrs7d / 7, 1) } else { 0 }
$correcaoPct = if ($totalPrs7d -gt 0) {
[math]::Round($correcaoCount * 100 / $totalPrs7d, 0)
} else { 0 }
$bloqueadorPct = if ($totalPrs7d -gt 0) {
[math]::Round($bloqueadorCount * 100 / $totalPrs7d, 0)
} else { 0 }
$top3 = @($prs7d | Where-Object { $_.tempo_spec_pr_min -ne $null } |
Sort-Object tempo_spec_pr_min | Select-Object -First 3)
$totalHistorico = @($state.prs).CountExibir relatorio no formato:
==================================================
[factory-metrics] {DATA} {HH:MM}
==================================================
METRICAS DOS ULTIMOS 7 DIAS
PRs de fabrica analisados: N
Tempo medio spec->PR: XX min
Tempo mediano spec->PR: XX min
PRs com correcao autonoma: N/N (XX%)
PRs com bloqueador humano: N/N (XX%)
PRs/dia (media 7d): X.X
TOP 3 mais rapidos:
#N titulo: XX min
...
ACUMULADO HISTORICO (desde inicio):
Total PRs registrados: N
==================================================Passo 5 -- Salvar state e agendar proxima iteracao
$state.ultimaExecucao = $agora.ToString('o')
$state | ConvertTo-Json -Depth 5 | Set-Content $stateFileEm seguida usar ScheduleWakeup:
- delaySeconds: 3600
- reason: "verificacao periodica do factory-metrics (coleta 1x/dia)"
- prompt:
<<autonomous-loop-dynamic>>
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.

