agentleFS
AgentShekelAgentShekel/agentic-workflow/skills/dev-methodology/references/agents.md

agentic-workflow / references

AGENTS.md4 starsChanged 29 days agoMITView on GitHub

Loaded by dev-methodology when an agent needs to identify which sibling agent to dispatch for a sub-task. Hot-path summary lives in dev-methodology/SKILL.md §"Agents" — it points here for the full per-role catalog. Agents are isolated subprocesses with fresh context. They receive input, do one job, return structured output.

The file

Files
agents.md

Dev Agents — full catalog

Loaded by dev-methodology when an agent needs to identify which sibling agent to dispatch for a sub-task. Hot-path summary lives in dev-methodology/SKILL.md §"Agents" — it points here for the full per-role catalog.

Agents are isolated subprocesses with fresh context. They receive input, do one job, return structured output.

Dev Track Leadership

  • dev-director — system-optimizer (out-of-band); does NOT accept engagements
  • dev-manager — acceptor for dev-lead handoffs; writes accept/reject verdict (does not plan or execute)
  • dev-lead — plans the engagement (the lead:plan step of the engagement-workflow Workflow); does NOT dispatch — the Workflow script fans out specialists/validators
  • dev-tech-architect — tech-spec author + task decomposition

Validators (run during spec/task creation)

  • userspec-quality-validator — document quality and completeness
  • userspec-adequacy-validator — solution feasibility
  • interview-completeness-checker — interview coverage gaps
  • tech-spec-validator — template compliance
  • skeptic — detects mirages (non-existent files/functions/APIs)
  • completeness-validator — bidirectional requirements traceability, over/underengineering, solution depth
  • task-validator — task template compliance
  • task-creator — generates task files from tech-spec
  • reality-checker — validates tasks against codebase
  • feasibility-assessor — validates feature research verdict (scope, risks, justification)

Reviewers (run during/after code writing)

  • code-reviewer — code quality across 10 dimensions
  • test-reviewer — test quality analysis with concrete fixes
  • security-auditor — OWASP Top 10, auth, input validation
  • performance-validator — N+1 queries, memory leaks, O(n²), async anti-patterns
  • migration-validator — DB migration safety: atomicity, rollback, data preservation
  • accessibility-validator — WCAG AA: ARIA, contrast, keyboard nav, screen readers
  • prompt-reviewer — prompt quality against prompt-engineering principles
  • documentation-reviewer — project-knowledge quality (haiku model — template check)
  • deploy-reviewer — CI/CD pipeline and deployment configuration (haiku model)
  • infrastructure-reviewer — folder structure, Docker, pre-commit hooks, .gitignore (haiku model)

Engineers (implementation)

  • dev-backend-engineer — server-side (APIs, services, DB, background jobs)
  • dev-frontend-engineer — client-side (UI components, state, routing)
  • dev-fullstack-engineer — end-to-end features when split adds no value
  • dev-devops-engineer — infrastructure + CI/CD
  • dev-technical-writer — project-knowledge maintenance
  • dev-qa-engineer — test strategy + plans

Research

  • code-researcher — codebase research for features (files, patterns, tests, integrations, risks)

QA

  • pre-deploy-qa — pre-deploy acceptance testing (tests + acceptance criteria)
  • post-deploy-qa — post-deploy verification on live environment (MCP tools, AVP)

Meta

  • skill-checker — validates skills against skill-authoring standards

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.