Skip to content

Latest commit

 

History

History
81 lines (53 loc) · 2.79 KB

File metadata and controls

81 lines (53 loc) · 2.79 KB

retort

Retort framework for multi-tool AI agent team orchestration, sync generation, and quality-gated workflows.

Project Context

  • Languages: javascript, yaml, markdown

  • Backend: node.js

  • ORM: none

  • Database: none

  • Messaging: none

  • Architecture: monolith

  • API Style: mixed

  • Default Branch: main

    • Phase: active

Coding Standards

  • Write minimal, focused diffs — change only what is necessary.
  • Maintain backwards compatibility; document breaking changes.
  • Every behavioral change must include tests.
  • Never commit secrets, API keys, or credentials. Use environment variables.
  • Prefer explicit error handling over silent failures.
  • Use the strongest type safety available for the language.
    • Conventional Commits (MANDATORY): All commit messages AND PR titles must use the format type(scope): description. Types: feat, fix, docs, style, refactor, test, chore, ci, perf, build, revert. Do NOT use natural-language titles like "Plan: Something" — CI will reject them.
    • Branch strategy: github-flow.
    • Code review: required-pr.
  • Generated file sync: After editing any file in .agentkit/spec/, you MUST run pnpm --dir .agentkit agentkit:sync and commit the regenerated output before pushing. CI drift checks will fail otherwise.

Authentication & Authorization

Provider: custom-jwt, strategy: jwt-bearer. Role-based access control (RBAC) is enforced — always check permissions before granting access to resources.

API Conventions

  • Versioning: url-segment
  • Pagination: cursor
  • Response format: envelope
  • Rate limiting is enabled

Testing

  • Unit: vitest

  • Integration: vitest

  • Coverage target: 80%

Always run the full test suite before creating a pull request. Never disable or skip existing tests without explicit justification.

Documentation

  • PRDs: docs/prd/

  • ADRs: docs/architecture/decisions/

  • API Spec: docs/api/

  • Technical Spec: docs/architecture/specs/

  • Brand Guide: .agentkit/spec/brand.yaml — AgentKit Forge (primary: #1976D2)

  • Quality Gates: QUALITY_GATES.md

  • Runbook: RUNBOOK_AI.md

Agent Teams

This project uses a multi-team orchestration model. Teams are specialized by domain:

  • Start with /orchestrate to assess the current state and coordinate work.
  • Use /plan to create structured implementation plans before coding.
    • Run /check to verify quality gates (lint, test, build) before committing.

See UNIFIED_AGENT_TEAMS.md for full team definitions and workflow phases. See COMMAND_GUIDE.md for when to choose /orchestrate, /plan, /project-review, and other commands.