Skip to content

security

diapason.skills.security

Skill security — capability validation and trust tiers.

TrustTier

Bases: str, Enum

Trust tier for a skill, ordered from most to least trusted.

classify_trust_tier(*, is_bundled=False, is_workspace=False, has_signature=False, in_index=False)

Return the trust tier for a skill based on its provenance.

Priority (highest to lowest): bundled > workspace > indexed > unreviewed.

has_dangerous_capabilities(manifest)

Return the subset of manifest's required capabilities that are dangerous.

validate_capabilities(manifest, allowed)

Return a list of capabilities required by manifest that are not in allowed.

An empty list means the manifest is fully authorized.