# Gaf Workflow Orchestrator

> Routes GAF requests to the correct specialized skill based on task phase, intent, and risk. Use when handling new automation requests, implementation changes, validation, deployment readiness, or governance checks.

- Skill: `miguel-carrera/gaf-workflow-orchestrator` (Agent Skill)
- Install (CLI): `npx skillmds@latest add miguel-carrera/gaf-workflow-orchestrator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/miguel-carrera/gaf-workflow-orchestrator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: miguel-carrera (https://skillmd.com/u/miguel-carrera)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/miguel-carrera/gaf-workflow-orchestrator

---


# GAF Workflow Orchestrator

## Goal
Select and apply the best GAF skill for the current user request while enforcing GAF process order.

## Router Rules
Always map intent to one primary skill first, then optionally a secondary skill.

### Phase 1: Intake and Documentation
- New task creation, setup, initial artifacts -> `gaf-task-bootstrapper`
- Scope unclear, likely large, needs decomposition -> `gaf-complexity-decomposer`
- Cross-file alignment review of docs/manifests -> `gaf-manifest-doc-consistency-auditor`
- Add/change Brain registry (catalog, storage map, dictionary, schemas) -> run `gaf-core-knowledge/scripts/` per `docs/brain-write-policy.md` (never edit governed YAML/SQL/MD directly); rule `gaf-brain-script-only-writes`

### Phase 2: Implementation
- Node runtime hardening and safe patterns -> `gaf-validation-guardrails-node`
- Writes/status/schema compliance -> `gaf-schema-quality-gate`
- Shared client/library change governance -> `gaf-shared-library-release-guard`
- Architecture boundary check (build-time vs runtime) -> `gaf-runtime-boundary-guard`

### Phase 3: Validation and Testing
- Formal final validation evidence -> `gaf-phase3-validation-runner`
- Final promotion decision support -> `gaf-predeploy-readiness-advisor` (then `node scripts/deploy-task.js <task_id>` from core-knowledge root; rule `gaf-deploy-script-enforcement`)

## Two-Skill Chains (Common)
- New task, then consistency:
  1) `gaf-task-bootstrapper`
  2) `gaf-manifest-doc-consistency-auditor`

- Implementation touching writes:
  1) `gaf-validation-guardrails-node`
  2) `gaf-schema-quality-gate`

- Predeployment checks:
  1) `gaf-phase3-validation-runner`
  2) `gaf-predeploy-readiness-advisor`

- Shared library release prep:
  1) `gaf-shared-library-release-guard`
  2) `gaf-runtime-boundary-guard` (if runtime behavior changed)

## Priority and Tie-Breakers
If multiple skills match, prioritize:
1. highest risk reduction (schema/runtime boundary/deploy blockers)
2. lifecycle order (Phase 1 before Phase 2 before Phase 3)
3. smallest change needed to unblock progress

## Safety Rules
- Do not skip required phase logic.
- Do not auto-promote to Production; use `deploy-task.js` only after explicit user confirmation.
- Governed Brain files: scripts only (`gaf-brain-script-only-writes`).
- If required inputs are missing, ask for them explicitly.
- Use TODO markers for unknown values; never invent governance-critical data.

## Output Contract
For each request, return:
- selected primary skill
- optional secondary skill
- why this routing was chosen
- immediate next action checklist (3-5 bullets max)