General-purpose trust router mapping verdict values to node names. Example:: graph.add_conditional_edges( "trust_check",

Wrap a LangChain tool with trust verification. Returns a callable that checks the agent's trust score before delegating to the original tool'

Remove an installed plugin. Args: name: Plugin name. Raises: MarketplaceError: If the plugin is not installe

* Extract an AI Card from an OCI manifest's layers.

Remove a plugin from the registry. Args: name: Plugin name. version: Specific version to remove. If ``None``, removes

Remove an agent from the process registry.

Remove a handler from all subscriptions.

Unsubscribe from a topic. Args: subscription_id: Subscription ID to unsubscribe

Restore original methods on a governed AutoGen agent. Removes all monkey-patches applied by :meth:`govern` and clears the agent from

Thread-safe update of the shared data store.

Update the resource usage for an agent

Update an agent's rate limit when their ring changes.

Update message status. Args: message_id: Message ID status: New status error: Optional e

Update the AI response for a specific turn. Useful when you add a turn with just the user message and want to update

Validate the dependency graph. Returns a list of validation errors, or empty list if valid.

Validate that a string contains none of the blocked keywords. Args: value: The text to scan (case-insensitive). metad

Validate the workflow definition. Ensures all transitions point to valid steps.

Validate the spec and return a list of errors.

* Validate code against a policy

Check if a challenge ID is valid and has not expired.

Validate agent input through Agent Shield Stage 1. Stage 1 screens the raw user message for jailbreak attempts, prompt injection, and

Verify a supervisor agent meets trust requirements. A supervisor at any level must declare a ``name`` and ``level``. Level-0 supervis

Verify signature (simulated)

* Verify integrity of all stored entries. * Returns false if any entry has been tampered with.

Verify multiple output pairs. Pure function that processes pairs in sequence. Args: outputs_a: Sequence of outputs from source

Verify the integrity of the full hash chain. Returns ``True`` if every entry's ``chain_hash`` matches its recomputed value and the ``

Verify a connection between two flow nodes.

Verify all agents in crew meet trust requirements. Returns: Dictionary with per-agent trust status and overall result.

Calculate drift between two probability distributions. Uses KL divergence and other statistical measures to compare distributions. Args

Verify a peer agent before IATP handshake. This is the core viral mechanism - unverified peers get directed to regis

Verify a Merkle inclusion proof.

Verify a provenance attestation against an artifact. Checks: 1. Artifact SHA-256 matches the attestation subject digest. 2. If a Sigstore

Verify trust for a pipeline stage.

Convenience property for tool version.

Block until a decision is made or timeout expires. Returns: The final decision, or ``PENDING`` if timeout was reached.

Pre-warm the JWKS cache for known partners. Per ADR-0003 200ms handshake SLA: cold-cache fetch is an HTTPS round-trip; pre-warming br

Poll the ATR source tree and recompile on any mtime change. Stdlib-only polling avoids adding a watchdog dependency. ``iterations`` bounds th

Serialise all entries and write them under a single lock. All lines are serialised before the lock is acquired, then written in a sin

Add member to sorted set with score.

Activate marks the kill switch active and records the event.

CanTransition reports whether a transition to the given state is allowed.

Clear marks the kill switch inactive and records the event.

Clear marks the given scope allowed again and records the event in registry history.

Clone returns a value-copy of the entry. Used at the AuditLogger API boundary so callers cannot mutate the in-store record (and thereby break the hash

Delete removes a credential. Returns true if it existed.

DestroySession force-removes the container for the given session.

Events returns a copy of the transition history.

ExportJSON serialises all audit entries to a JSON string.

ExtractSQLFacets is the built-in SQL extractor. Reads sub["query"] and returns {verb, target, tables, functions} as flat strings.

GovernOperation executes a single operation behind the standard governance stack.

InjectEnv renders placeholders in a subprocess environment map.

ListHandles returns all credential handle names.

LoadCedar registers a Cedar backend with the policy engine.

NewCredentialInjector constructs an injector backed by the given vault.

NewKillSwitch creates an inactive kill switch.

Placeholder returns the {{cred:NAME}} placeholder for this handle.

ScanAll scans every tool in the slice and returns per-tool results.

ScanConfigPaths scans directories for agent configs, containers, and dependency markers.

ScanProcesses returns merged discovered agents from caller-supplied process metadata.

WithTrustConfig overrides the default trust configuration.

Attach rate limit headers to the response.

Add AgentMesh verification footer to tool output.

Analyze content to identify important factors.

Analyze common section patterns across documents.

Append a single JSON line to the file.

Apply time-based score recovery toward neutral (0.7).

Apply the resolution to escrow and reputation.