affaan-m/ecc

platform-factory

Build deterministic OCI images, SBOM/provenance evidence and hardened container or microVM runs from a compiled executable with the Platform Factory (pf) CLI, and keep its freeze lock in sync with a companion Claude Code hook.

Voir la source

Mode d’emploi

  1. Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
  2. Consultez le document original pour les outils, identifiants et environnements requis.
  3. Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/affaan-m/ecc --skill platform-factory

Avant de commencer

  • L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
  • Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Document Skill original

Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.

Platform Factory

Platform Factory (platform-factory, alias pf) is a single Go binary that turns a compiled executable into a deterministic OCI image, attaches native supply-chain evidence (SBOM, provenance, signatures, policy) and runs the result as a hardened container or a KVM/HVF microVM. No Docker daemon, BuildKit, syft or cosign is needed for the core path. It plugs into an existing stack; it does not replace Kubernetes, CI orchestrators or registries.

Full reference: the project's CLI guide. This skill covers the workflow an agent should follow and the guard hook that protects it.

When to Activate

  • The repository has a .platform-factory/ directory or freeze.lock.json.
  • The user runs, or asks about, pf / platform-factory commands.
  • The task is reproducible OCI image builds, SBOM/provenance evidence, or microVM execution of a legacy or dynamically linked binary.
  • You edit a dependency manifest (package-lock.json, go.sum, requirements.txt, Cargo.lock, pom.xml, ...) in a Platform Factory project.

Core Workflow

bash
pf detect ./my-app        # native executable inspection; JSON output
# Directory detection runs loaded language-plugin inspection commands.
pf init --language node .    # create pf.yaml (language plugin must be loaded: pf plugin load node)
pf plan                   # project plan (platform-factory.dev/project-plan/v1)
pf project freeze         # pin dependencies -> .platform-factory/freeze.lock.json
pf build                  # deterministic OCI image
pf verify .platform-factory/image   # strict release gate: digests, diff_ids, platforms
pf run --runtime podman .platform-factory/image
pf diff                   # compare builds

pf launch --publish --yes freezes when required, builds, verifies, publishes to a registry and then starts the workload in one step, so it is an outward-facing action (see rule 7). pf sbom PATH... produces a reproducible SBOM natively.

Rules that matter

  1. Freeze before build. Any change to a dependency manifest makes .platform-factory/freeze.lock.json stale. Run pf project freeze, then pf build.
  2. Treat `pf verify` as the gate. A build is not done until verify passes.
  3. Inspect plugins before directory detection. Native executable classification does not run the target, but directory detection invokes loaded language plugins. Review those plugins and project commands as executable configuration before detect, plan, freeze, or build.
  4. Ambiguity is an error. pf detect exits with code 2 when a directory matches several ecosystems. Pass --accept-ambiguous only when the user confirmed the choice.
  5. Never add secrets to build inputs. Inputs end up in the image and its evidence.
  6. Callbacks are opt-in. pf hooks (CLI/MCP callbacks) only run when PLATFORM_FACTORY_HOOKS_CONFIG points to an explicit absolute file, so a cloned repository cannot activate executables.
  7. Keep execution within the user's authorization. Before publishing, deploying, launching, or running a workload, identify its target, image reference, and effects. Reuse explicit authorization already given for that scope; ask only when the action exceeds it. Use --dry-run where available before the authorized mutation. Do not add --yes to bypass an approval the user has not given.

Companion Claude Code Hook: freeze guard

The project ships a Claude Code plugin, `platform-factory-guard`, that runs next to ECC's own hooks:

EventTriggerBehavior
PostToolUseEdit/Write/MultiEdit on a dependency manifestReminds you to run pf project freeze
PreToolUseBash running `pf build\publish\launch\deploy`Warns when a manifest is newer than the freeze lock; denies if PF_GUARD_STRICT=1

It never executes pf, recognizes project configuration such as pf.yaml as well as .platform-factory/, ignores the per-user plugin directory as a project marker, and fails open. Disable with PF_GUARD=off.

Install it as a plugin (requires Node.js 18+):

text
/plugin marketplace add CYPT71/platform-factory
/plugin install platform-factory-guard@platform-factory

Anti-Patterns

bash
# BAD: build right after changing dependencies
npm install left-pad && pf build

# GOOD: refresh the lock first
npm install left-pad && pf project freeze && pf build && pf verify .platform-factory/image
  • Skipping pf verify and shipping the image.
  • Passing --accept-ambiguous to silence exit code 2 without checking which ecosystem is intended.
  • Editing .platform-factory/freeze.lock.json by hand instead of re-running pf project freeze.
  • Running pf with a PLATFORM_FACTORY_HOOKS_CONFIG file you did not review.

Best Practices

  • Build the executable statically when possible (CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -trimpath -ldflags='-s -w') so the image is small and reproducible.
  • Select a container or microVM from the workload's isolation requirements and available host support. This external CLI guide does not certify an ECC sandbox tier or replace ECC capability/evidence policy.
  • Use pf extensions list / pf extensions analyze for offline policy checks (AI, data, gateway, Kubernetes, supply-chain analyzers).
  • In CI, run pf verify as a required step and archive the SBOM and provenance.

Related Skills

  • docker-patterns - container conventions when a Dockerfile is still involved
  • deployment-patterns - rollout and release strategies
  • kubernetes-patterns - running the produced image on a cluster
  • security-review - reviewing supply-chain and secret handling
du même dépôt

Autres Skills

Tous les Skills
affaan-m
Communauté

rust-testing

Rust testing patterns including unit tests, integration tests, async testing, property-based testing, mocking, and coverage. Follows TDD methodology. Use when writing Rust tests — unit, integration, async, property-based, or coverage.

Installations / 24h
489
GitHub Stars
308,2 k
Mis à jour
20 nov.
affaan-m
Communauté

search-first

Research-before-coding workflow: search npm/PyPI, MCP servers, skills, and GitHub for existing tools before writing custom code, then adopt, extend, or build. Launches the researcher agent for non-trivial needs. Use when starting a feature, adding a dependency or integration, or about to write a utility that may already exist.

Installations / 1h
5
GitHub Stars
308,2 k
Mis à jour
20 nov.
affaan-m
Communauté

browser-qa

Run automated post-deploy UI verification with a browser automation MCP (claude-in-chrome, Playwright, or Puppeteer): console-error and Core Web Vitals smoke checks, form and auth-flow interaction tests, screenshot visual regression across three breakpoints, and axe-core accessibility audits ending in a SHIP / DO-NOT-SHIP verdict. Use when testing a deployed feature on staging or preview, before shipping frontend changes, reviewing a frontend PR, or checking responsive layout and accessibility.

Installations / 1h
2
GitHub Stars
307,9 k
Mis à jour
20 nov.
affaan-m
Communauté

docker-patterns

Docker and Docker Compose patterns for local development, hardened CLI installer harnesses, container security, networking, volumes, and multi-service orchestration. Use when creating or reviewing Dockerfiles and Compose services, testing installers across Linux distributions, or planning accurate native macOS and Windows validation.

Installations / 1h
2
GitHub Stars
307,9 k
Mis à jour
20 nov.