Append a legacy flat rule.

Add an agent to a workflow. Args: workflow_id: Workflow ID agent_id: Agent to add depend

Add an ethical alignment rule. Args: principle: Which principle this enforces description: What the

Add a custom policy rule. Args: rule: Dictionary defining the policy rule with keys: - name: Rule name

Record a dependency relationship between two SPDX IDs.

Add a document to the corpus.

Add an edge to the graph

Add a maintenance window to the temporal graph

Add a manually discovered vulnerability (not from OSV).

Add a policy constraint to the policy graph

Add a policy rule to the graph

Register a cross-namespace communication rule. Args: rule: The NamespaceRule to add.

Add or update a server in the allowlist.

Add a risk signal for an agent. Critical signals trigger an immediate score recalculation. Args: agent_did: The agent's

Add SLO/SLI metrics data to the dossier. Args: data: Dictionary of SLO metric names to values.

Add transaction to mempool

Add AgentMesh verification footer to content.

Post-run hook: drift detection on result. Args: result: The agent run result. **kwargs: Additional run context.

Final audit summary and context teardown.

Post-tool hook: audit the tool execution result. Args: tool_name: Name of the tool that was called. result: The tool'

Snapshot for the ``agent_shutdown`` hook. When ``tool_calls``, ``tokens`` or ``duration_seconds`` are not supplied the builder defaul

Snapshot for the ``agent_shutdown`` intervention point (§2.8).

Snapshot for the ``agent_startup`` intervention point (§2.1).

Find agent IDs in a given state.

Aggregate multiple embedding verification scores with profile context. Args: scores: Sequence of VerificationScore objects t

Aggregate results into summary statistics.

Aggregate multiple verification scores into summary statistics. Pure function. Args: scores: Sequence of VerificationScore obje

* Convert an AI Card back to an AGT agent identity.

Get total persisted alert count.

Get burn rate alerts (single 24h window).

Check both per-agent and global limits. Returns True if allowed.

* Allow a specific violation once

Analyze document structure and return insights. Returns: Analysis results including section importance and relat

Analyze contract for risky clauses.

Analyze the entire corpus to identify patterns. Returns: Analysis results including common patterns and optimiza

Analyze an image with a text prompt. Args: image: Image to analyze prompt: Text prompt for analysis

AgentMesh - The Secure Nervous System for Cloud-Native Agent Ecosystems Identity · Trust · Reward · Governance

Apply time-based decay to an agent's trust score.

Approve a pending tool call by its call_id.

Archive the session after audit commitment.

Serialise stats to a plain dictionary.

* Cursor-specific: Ask Cursor AI for a safe alternative to blocked code

Assemble buffered SSE bytes into a chat-completion response. Fails closed on malformed framing, multi-choice streams, unsupported fields, or

Governed async streaming execution.

TODO: Create a TrustHandshake from initiator → responder and execute it. Print: [OK] Handshake succeeded — session_token: <token>

Audit claim against observation

Audit the wisdom database against a new model. Args: new_model: The new model to test with (e.g., "gpt-4o", "gpt

Authorize an MCP tool call. Checks (in order): 1. Agent DID is present (if required) 2. Agent DID is not blocked

Ingest multiple signals in batch. Args: raw_signals: List of raw signal data Returns:

buildContentsAPIPath assembles the GitHub contents-API path for a given repo and file path, URL-escaping each segment so that values containing `?`, `

The user's existing AGT policy. Untouched by SpendGuard composition.

Safely evaluate a mathematical expression using AST parsing.

calculateTrustScore calculates the trust score based on the manifest

Calculate aggregate metrics from results

Calculate fault impact score — simple pass/fail based on success rate.

Return True only if the agent passes all critical compliance checks.