fix(policy): PolicyContext.trust_level hardcoded to Trusted — trust_level rule condition non-functional #2112
Description
Problem
In src/runner.rs:746, the PolicyContext passed to PolicyGateExecutor always has trust_level = TrustLevel::Trusted (hardcoded). The trust_level field in policy rules has no effect because the context always reports the highest trust level.
Impact
Impossible to write a policy rule that restricts tool access based on actual skill trust tier (quarantined vs verified vs trusted).
Expected behavior
PolicyContext.trust_level should reflect the actual trust level of the invoking skill. Fall back to TrustLevel::Trusted for direct user calls.
Notes
- Feature is cfg(feature = policy-enforcer) and enabled = false in default config — no production impact today
- Found during static code analysis in CI-71 (2026-03-22)
- Related: crates/zeph-tools/src/policy.rs:175-181, src/runner.rs:746