@a5c-ai/policy-adapter

Proof-based policy enforcement trust core: unified signed envelopes, trusted-store key resolution, config-integrity manifest, and the proven bridge.

This package is a runtime dependency of @a5c-ai/comm-adapter, @a5c-ai/tools-adapter and @a5c-ai/transport-adapter, so it is published alongside them. It is a library, not a CLI — there is no binary and no standalone entrypoint.

Install

npm install @a5c-ai/policy-adapter

Entrypoints

| Export | Module | Purpose | | --- | --- | --- | | @a5c-ai/policy-adapter | dist/index.js | Policy evaluation, authorization issue/store, exec-seam registry, argv matching and arg canonicalization. | | @a5c-ai/policy-adapter/trust | dist/verify-envelope-trusted.js | Verify a signed envelope against the trusted key store. | | @a5c-ai/policy-adapter/config-manifest | dist/config-manifest.js | Build and check the config-integrity manifest. | | @a5c-ai/policy-adapter/proven-bridge | dist/proven-bridge.js | The proven bridge between an evaluated policy decision and the enforcing seam. |

Scope

The package holds the decision and verification primitives only. Wiring them onto a production path (which exec seams are gated, which credentials identify a caller) is the consumer's job — see policy-enforcement-wiring in the main entrypoint for the seam that consumers call.

Signing and envelope primitives live one layer down in @a5c-ai/trust-core.

License

MIT