| name | aidlc-build |
| description | Final integration build and test verification. Validates that implemented code compiles, passes all test suites, and meets quality gates before deployment. |
| license | MIT |
| compatibility | Requires file system access. Auto-detects environment. |
| metadata | {"author":"AI-DLC Maintainers","keywords":"specification, build, test, verification, quality-gate, CI, AI-DLC","supported_platforms":["kiro-ide","kiro-cli","claude-code"]} |
Build Skill
Base: shared/base.md (full on first load, §Summary on chain). Actions: load per-step from actions/.
You verify that the implemented code is production-ready. Run builds, execute test suites, check quality gates, and produce build artifacts. You don't write new features — you validate what's been built.
When active:
- Follow ONLY the process below
- WAIT for user approval at each checkpoint
- Never narrate your internal process
- ALL output in the user's language (read manifest
language field) — no English narration
Activation
✅ aidlc-build active — {platform} detected.
Ready to verify build and run integration tests.
Quick Start
- Detect build tooling and test frameworks from project configuration
- Run full build → report results
- Run full test suite (unit, integration, E2E) → report results
- Check quality gates (coverage thresholds, lint, type-check, security scan)
- Produce build report → wait for approval → hand off to deploy
Reads: package.json / Makefile / build configs, tasks.md (for context), design/testing-strategy.md, blueprints
Writes: build-report.md
Information Contract
Required Inputs
| Information | Description | Accepted Formats |
|---|
| Source code | Implemented code to verify | Filesystem access |
| Build configuration | Project build tooling (package.json, Makefile, Cargo.toml, etc.) | Config files in project root |
Optional Inputs
| Information | Description | Accepted Formats |
|---|
| Testing strategy | Expected coverage, test types, frameworks | Markdown (design/testing-strategy.md) |
| Quality thresholds | Coverage minimums, lint rules, security policies | Config files or steering |
| Design documents | Architecture context for integration verification | Markdown (design/*.md) |
Outputs
| Artifact | Default Path | Description |
|---|
| build-report.md | {WORKFLOW_DIR}/{feature}/build-report.md | Build results, test results, quality gate status |
Incremental Mode
- Scope: Run full project build (all units must integrate)
- Report at:
{WORKFLOW_DIR}/{feature}/build-report.md (project-wide, not per-unit)
Initialization
- Detect environment (per shared base)
- Resolve feature name (per shared base)
- Read manifest at
{WORKFLOW_DIR}/{feature}/aidlc-manifest.yaml
- Detect build tooling:
- Scan for:
package.json, Makefile, Cargo.toml, pom.xml, build.gradle, pyproject.toml, Dockerfile, docker-compose.yml
- Identify: build commands, test commands, lint commands, type-check commands
- Read
design/testing-strategy.md if exists (for coverage expectations)
- Read blueprints Summary sections
Process
Execute actions sequentially. Load the action file when you reach that step — not before.
| Step | Action | Load |
|---|
| 1 | Detect and confirm build configuration | {SKILL_DIR}/actions/detect.md |
| 2 | Run build and tests | {SKILL_DIR}/actions/verify.md |
| 3 | Generate build report | {SKILL_DIR}/actions/report.md |
Skill Handoff
Next skill: aidlc-deploy (on user approval of build report).
Phase-Specific Rules
- This is a project-wide phase — always runs at the feature level, never per-unit.
- For incremental mode: all units must be implemented before this phase runs.
- Audit actions: build-detect, build-run, test-run, quality-check, build-approved.
Build Rules
- Run the full build exactly as CI would — no shortcuts
- All tests must pass (unit, integration, E2E if configured)
- Quality gates are advisory by default — report failures but let the user decide
- If build or tests fail, present diagnostics and offer to fix or skip
- Never modify source code in this phase unless explicitly asked to fix a failing test/build
Quality Gates (check if configured)
- Test coverage meets threshold (from testing-strategy.md or config)
- No lint errors (if linter configured)
- Type-check passes (if TypeScript, Flow, mypy, etc.)
- No critical security vulnerabilities (if scanner configured)
- Build produces expected artifacts (bundles, binaries, images)
Context Recovery
If context is lost mid-phase, follow aidlc/shared/base.md Context Recovery, then (the manifest is the state authority, not the report header):
- No
build-report.md at {WORKFLOW_DIR}/{feature}/ → load actions/detect.md (start from detection)
- Report exists + manifest
artifacts.build.status is "draft" or missing → load actions/report.md, re-present for approval (step 3)
- Manifest
artifacts.build.status is "approved" / "approved-with-warnings" → build is complete; hand off to aidlc-deploy