ll-qa-api-from-feat

name	ll-qa-api-from-feat
description	ADR-053 unified entry skill for running the complete API test chain (feat→apiplan→manifest→spec) with acceptance traceability. Orchestrates sub-skills via Skill tool.
version	1.0
adr	ADR-047, ADR-052, ADR-053
category	qa
chain	api-test
phase	full-chain
supersedes	[]

LL QA API From FEAT

One-shot skill that runs the complete API test chain from a frozen FEAT document.

Input

• feat_freeze_package: directory path from ll-product-epic-to-feat (must be frozen)
• feat_ref: the specific FEAT identifier

Output

• ssot/tests/api/{feat_id}/api-test-plan.md — Contains Acceptance Traceability table
• ssot/tests/api/{feat_id}/api-coverage-manifest.yaml — Generated by ll-qa-api-manifest-init
• ssot/tests/api/{feat_id}/api-test-spec/SPEC-*.md — Generated by ll-qa-api-spec-gen

Internal Pipeline

1. ll-qa-feat-to-apiplan — FEAT → api-test-plan (with Acceptance Traceability)
2. ll-qa-api-manifest-init — api-test-plan → api-coverage-manifest
3. ll-qa-api-spec-gen — api-coverage-manifest → api-test-specs

Options

• --preview: Only run up to api-test-plan, stop before manifest-init
• --no-spec: Stop after api-coverage-manifest, skip spec-gen
• --target P0|P1|P2: Filter by priority

Orchestrator Behavior

State Machine

START → APIPLAN_RUNNING → TRACEABILITY_VALIDATING → MANIFEST_RUNNING → SPEC_RUNNING → COMPLETE

• START: Validate feat_ref and feat_freeze_package status=frozen
• APIPLAN_RUNNING: Call Skill tool /ll-qa-feat-to-apiplan with feat_freeze_package + feat_ref
• TRACEABILITY_VALIDATING: Validate Acceptance Traceability table completeness
• MANIFEST_RUNNING: Call Skill tool /ll-qa-api-manifest-init with api-test-plan output path
• SPEC_RUNNING: Call Skill tool /ll-qa-api-spec-gen with api-coverage-manifest output path
• COMPLETE: Return execution summary with paths to all generated artifacts

Error Handling Strategy

error_handling:
  strategy: fail-fast
  failure_at:
    apiplan: abort
    manifest_init: abort
    spec_gen: continue  # preserve manifest on spec_gen failure
  artifacts_on_failure: keep
  retry: disabled

Acceptance Traceability Validation

validation:
  acceptance_traceability:
    check_mandatory_mapping: true     # each capability must have AC mapping
    check_complete_coverage: true     # each AC must have at least one capability coverage
    fail_on_incomplete: true
    validation_point: before_manifest_init

Non-Negotiable Rules

• Same as ll-qa-feat-to-apiplan (input validation, acceptance traceability)
• Same as ll-qa-api-manifest-init (cut rules, four-dimensional status)
• Same as ll-qa-api-spec-gen (spec completeness)
• Acceptance traceability table must be complete before proceeding to manifest-init
• Orchestrator does not directly generate artifacts — only calls sub-skills via Skill tool

Evidence

• Execution evidence: evidence/execution-evidence.schema.json
• Supervision evidence: evidence/supervision-evidence.schema.json