convez.ai
AgenteDevOps & InfraestruturaAvançado
nathanjohnpayne

Mergepath

nathanjohnpayne/mergepath

Padrão de repositório para alinhar agentes IA e humanos no desenvolvimento

Ver no GitHub
BomPluginScore · de 0 a 100

Padrão determinístico de repositório que mantém humanos e agentes de IA alinhados durante desenvolvimento colaborativo. Implementa um sistema de controle de mudanças com política de revisão automática, instruções para agentes e estrutura de governança para múltiplas ferramentas operarem sem drift de configuração.

Instalação

$npx convez-ai nathanjohnpayne/mergepath

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
5
Forks
0
Acompanhando
0
Saúde
71%
Último commit
há 2 dias
Versão
—
Issues abertas
765
Contribuidores
2

Casos de uso

  • Governança de repositórios multi-agente com políticas automatizadas
  • Padronização de fluxos de trabalho entre agentes de IA e ferramentas de desenvolvimento
  • Implementação de review policies determinísticas com fallbacks manuais
  • Orquestração de deploy com credenciais 1Password e Firebase

Veredito

Mergepath é uma arquitetura sólida para equipes que trabalham com múltiplos agentes de IA em repositórios complexos. Exige compreensão profunda de CI/CD, controle de acesso e fluxos de aprovação — adequado para contextos onde consistência determinística entre agentes é crítica.

README

A tradução para o português deste README ainda está na fila.

Mergepath

Reference implementation of the AI Agent Tooling Standard.

The goal is to allow multiple AI coding agents and development tools to operate consistently without configuration drift. See BRAND.md for the umbrella vocabulary (Playground, Cockpit, Tiebreaker, Checks) and naming history.

For AI Agents

Read these files in order before taking any action:

  1. AGENTS.md — behavioral instructions and operating rules
  2. rules/repo_rules.md — binding structural constraints
  3. Relevant specs/ files — intended behavior
  4. .ai_context.md — supplemental system context

Code Review Policy

Every change in this repository goes through the policy in REVIEW_POLICY.md, including a self-peer review by the authoring agent's reviewer identity and, for changes that cross the threshold or touch protected paths, automated external review via the OpenAI Codex GitHub app (Phase 4a) or a manual CLI fallback (Phase 4b).

Key Files

FilePurpose
AGENTS.mdInstructions for AI agents
DEPLOYMENT.mdBuild and deployment
CONTRIBUTING.mdDevelopment workflow
.ai_context.mdHigh-level system context
BRAND.mdMergepath umbrella vocabulary (surfaces, reserved names, naming history)
mergepath/playground/index.htmlMergepath Playground — tune the review policy and replay recent PRs against the draft
scripts/policy-sim.shBakes real gh PR data into a temp copy of the Mergepath Playground for local replay
ai_agent_tooling_standard.mdFull repository standard (reference)

Firebase Auth Template

This template includes the canonical Google Cloud and Firebase helper scripts for this account:

  • scripts/gcloud/gcloud installs a local wrapper so ordinary gcloud commands can use 1Password-backed or explicit source credentials without a routine interactive gcloud auth login, while attributing quota to the resolved target project from explicit flags, the repo's .firebaserc, or the active gcloud config.
  • scripts/firebase/op-firebase-setup creates a per-project firebase-deployer@{project-id}.iam.gserviceaccount.com, grants deploy roles, and configures impersonation.
  • scripts/firebase/op-firebase-deploy resolves a source credential per the canonical precedence (project Firebase-vault SA key first, with shared 1Password ADC and impersonation as fallbacks) and runs firebase deploy with the target project stamped in as the quota project.

The canonical source-credential precedence is documented in DEPLOYMENT.md § Deploy credential precedence (canonical). The default day-to-day credential — interactive and CI — is the per-project Firebase-vault SA key (op://Firebase/{project-id} — Firebase Deployer SA Key), with the shared 1Password ADC as a fallback and an explicit GOOGLE_APPLICATION_CREDENTIALS as the highest-priority override.

This 1Password-first deploy-auth model is intentional. Do not revert template-derived repos to deploy-key-on-disk or routine firebase login-based guidance unless a human explicitly requests that change.

See DEPLOYMENT.md for the full bootstrap and deploy flow.

Directory Structure

DirectoryPurpose
rules/Binding repository constraints
specs/Intended system behavior
plans/Execution and migration plans
mergepath/Mergepath Playground and reserved slots for future Mergepath surfaces (see BRAND.md)
tests/Automated validation
src/Application code
functions/Backend handlers
scripts/Build, CI, and automation tooling
docs/Architecture and design documentation
dist/Generated build artifacts (do not edit manually)
Ver o original no GitHub

Bom para

Relacionados