Struct PolicyDecisionPoint 

pub struct PolicyDecisionPoint { /* private fields */ }

The local policy decision point. Evaluates tool calls against the intent map, the global mission, and every per-agent mission that selects the caller.

Implementations

impl PolicyDecisionPoint

pub fn from_config( config: &HarnessConfig, signer: Option<Arc<TokenSigner>>, ) -> Self

Build from harness config with an optional token signer.

pub fn applicable_missions<'s>( &'s self, caller: &CallerFacts<'_>, ) -> Vec<&'s MissionConfig>

Every mission that constrains caller: the global mission, then each [[missions]] entry whose selectors match.

pub fn decide(&self, tool: &str, now_s: u64) -> AuthzDecision

Decide whether tool may run at now_s for a caller without a validated identity. Signs nothing.

pub fn decide_for( &self, tool: &str, now_s: u64, caller: &CallerFacts<'_>, ) -> AuthzDecision

Decide whether tool may run at now_s for caller. Signs nothing, so the pipeline can run it before the budget is charged. Missions only narrow: every applicable mission must allow the call.

pub fn intent_for(&self, tool: &str) -> String

The business intent tool maps to: its intent_map entry, or the tool name itself. A plain lookup that never denies.

pub fn mint_intent_token( &self, tool: &str, intent: &str, caller: &CallerFacts<'_>, audience: &str, now_s: u64, ) -> Option<String>

Sign a short-lived intent token for a permitted call, bound to the backend that receives it. sub is the human principal and act.sub the calling agent; the token never outlives an applicable mission. None when no signer is configured.

pub fn intent_token_ttl_s(&self) -> u64

The configured intent token TTL.