Gaf Task Bootstrapper

GAF Task Bootstrapper

Goal

Create a consistent gaf-automations-library/Test/{task_id} scaffold that matches GAF core-knowledge conventions.

Use This Skill When

• User asks to create a new automation task.
• User asks to initialize task files from templates.
• Work is at Phase 1 (Intake and Documentation) or early Phase 2.

Required Inputs

• task_id (kebab-case)
• task_name
• owner
• complexity_tier (SIMPLE, MID_COMPLEX, or COMPLEX)
• primary_storage_dependencies
• shared_dependencies
• runtime (node default unless explicitly requested otherwise)

Workflow

1. Register in catalog via node scripts/register-task.js <payload.json> from gaf-core-knowledge/ (status: DRAFT, environment: test — location defaults to Test/{task_id}). Read task-catalog.yaml; do not edit by hand.
2. If storage entities are missing, add them via Brain scripts (register-storage-node.js, register-entity.js, update-db-schema.js per docs/brain-write-policy.md).
3. Create gaf-automations-library/Test/{task_id}.
4. Create and align:
   • task-definition.md
   • manifest.yaml
   • automation-doc.md
   • runtime entrypoint run.js (Node default and required convention)
   • scripts/ folder for internal subtask/module files invoked by run.js
   • package.json (if runtime needs local deps/scripts)
5. Ensure identifiers are synchronized across all files:
   • task_id
   • status
   • owner
   • version
6. Add TODO markers for unknown values instead of inventing fields.

Guardrails

• Do not invent undeclared storage nodes/entities.
• Keep runtime self-contained; no runtime reads from gaf-core-knowledge.
• Reference published shared libraries (*_v<n.n.n>) in manifest.yaml for deploy-ready tasks.
• Follow Test-first flow; do not auto-promote to Production (deploy-task.js only after validation and user confirmation).

Output Format

Return:

• files created
• unresolved TODOs requiring user input
• quick checklist for next step (Phase 2 or Phase 3)