Skip to content
DEVKIT INDEX

Skill definition

hi-tech-build-audit

Audit a repository's technology stack, build system, dependency management, CI/CD pipelines, deployment targets, and platform assumptions. Use for build onboarding, migration readiness, deployment review, or CI/CD documentation.

Documentation

SKILL.md

Technology and Build Audit

Read shared/leaf-contract.md. Produce verified build and deployment evidence. Do not own module reconnaissance or cross-artifact synthesis.

Inputs

Require a readable absolute repo_root. Accept target_env: local, container, cloud, hybrid, or auto (default). Accept depth: quick, standard (default), or deep. Default output to discovery-output/audit-data.

Workflow

  1. Validate input and output boundaries.
  2. Use retrieval rules in audit-checklist.md and shared/retrieval-protocol.md. Discover callable metadata; fast-fail unavailable layers once.
  3. Inventory languages, runtimes, frameworks, package managers, build systems, dependency locks, generated code, and version constraints.
  4. Verify build, test, lint, packaging, and local-run commands from files. Label convention-inferred commands; never present them as verified.
  5. Inspect CI/CD triggers, jobs, gates, artifacts, secrets references, environments, and deployment targets.
  6. For deep, trace API-layer and runtime dependencies only where graph/source evidence establishes them.
  7. Redact values of secrets and .env entries while preserving key names and evidence locations.
  8. Write artifacts per output-contract.md and validate required outputs.

Outputs

  • build-report.md, tech-stack.json, ci-cd-pipelines.md, and artifact-manifest.json — required.
  • platform-targets.md — when deployment targets detected.
  • api-dependencies.md — when depth=deep.

Complete only when each claimed command or technology cites repository evidence, required artifacts validate, detected/absent/unknown are distinct, and the manifest reflects partial or blocked evidence honestly.

Return a prerequisite or evidence-gap result to the root when another capability is needed.

Continue exploring

Related skills

hi-behavior-modelingProduce draft-only evidence-backed technical behavior as sequence, activity, state-machine, and event-flow documentation. Do not use for business use-case catalogs, API contracts, or persistent data models.hi-data-model-discoveryProduce draft-only evidence-backed physical and logical data models, ERDs, entity and field inventories, data dictionaries, ownership, read/write access, lifecycle, sensitivity, caching, replication, and migration gaps. Do not execute schema migrations or present inferred relationships as declared constraints.hi-docs-seekerFind current, authoritative technical documentation for libraries, frameworks, APIs, tools, standards, versions, migrations, configuration, and compatibility. Use when an answer requires verified external documentation or precise source links.