Expand description
proof-policy: minimal declarative policy engine (ARCHITECTURE §5). Implicit AND over a closed requirement set. No OR/NOT, no code evaluation, no network. Same evidence under different policies may yield different decisions: Evidence ≠ Policy, Proof ≠ Trust decision.
Re-exports§
pub use eval::evaluate_policy;pub use eval::EvalInputs;pub use eval::PolicyOutcome;pub use eval::RequirementResult;pub use eval::RevocationSet;pub use explain::explain_full;pub use explain::explain_outcome;pub use explain::explain_report;pub use policy::canonical_policy_hash;pub use policy::parse_policy;pub use policy::policy_to_canonical_cbor;pub use policy::Policy;pub use policy::Requirement;pub use state::state_from_report_and_proof;pub use state::VerifiedState;
Modules§
- eval
- Deterministic policy evaluator (implicit AND). Pure function of (state, policy, explicit trust inputs) — no I/O, no crypto, no AI.
- explain
- Evidence-derived explanations. Pure projections of report + outcome: the text can never contradict the machine verdict (tested on every vector).
- policy
- Policy schema + validator (spec §9). JSON is the input language; it is validated BEFORE anything is evaluated, and never executed as code.
- state
- VerifiedState: the exact, minimal view of a verified proof that policy
evaluation may consume. Derived from a
VerifyReportplus the parsedProofit describes — never from databases, reputation, or prose.