Main election and heartbeat loop

Classify a single Cargo dependency and emit a finding if needed.

Emit governance data as Agent-Lightning spans.

Emit reward to Agent-Lightning.

Encrypt with ChaCha20-Poly1305.

Encrypt data with AES-256-GCM. Requires the ``cryptography`` package (``pip install cryptography``). Falls back to a clearly-marked n

Enforce the global ``require_approval`` set computed at construction time. Returns a denial dict when the action is globally protected

Bound verification attempts per actor in a sliding window.

Enrich a single section with metadata prefix. Args: section: Section to enrich document: Parent docu

Ensure the Docker image is available, pulling if necessary.

Ensure all parent directories exist for a given path.

Create the sandbox group on first use when configured. Probes the mgmt client for the appropriate get/create methods — the exact surf

Build the v5 :class:`ExecutionContext` on first need.

Build the v5 :class:`ExecutionContext` on first need.

Lazily start the background worker on first sink registration.

Acquire the shadow lock and (if outermost) replace blocked modules. Safe to call recursively from the same thread; other threads block on the

Estimate text complexity based on various factors.

Evaluate a single Condition against the context.

Evaluate a single condition — mirrors SharedPolicyEvaluator logic.

Evaluate a simple policy expression. Supports the same syntax as ``PolicyRule._eval_expression``: - ``field == 'value'`` - ``field`` (boo

Evaluate a simple expression.

Evaluate via cedarpy Python bindings.

Evaluate via cedarpy Python bindings.

Evaluate via cedar CLI subprocess.

Evaluate a single policy condition against a context dict.

Evaluate using local `opa eval` CLI. When the OPA CLI is not available, returns a deny decision with an explicit error instead of sil

Evaluate a single manifest against policy rules.

Mock Cedar pattern evaluator for testing/dev only. Does NOT enforce principal or resource constraints. Use mode='cedarpy' or mode='cl

Mock Rego evaluator for testing/dev only. Supports simple ``==``, ``!=``, and ``not`` conditions. Does NOT support comprehensions, se

Mock Cedar pattern evaluator for testing/dev only. Parses simple Cedar policy patterns: - permit(principal, action == Action::"X",

Run `opa eval` subprocess.

Evaluate a single multi-agent policy.

Query a remote OPA server via REST API.

Evaluate how well the agent responded to a query. Returns a score from 0 to 1.

Evaluate policy rules against context. Args: context: Policy evaluation context Returns: Action str

Evaluate a merged rule list from folder-scoped discovery.

Evaluate using folder-level policy discovery and merge.

Return trust score for the given agent. Verifies the agent's Ed25519 signature over its DID. The agent must sign the DID string with

Run the trust evaluator under the read side of the RW lock.

Run the evaluator under the read side of the RW lock.

A fresh SSE response start without stale content-length/encoding.

Remove expired entries from the verified clients cache.

Execute the action (stub - real implementation dispatches to handlers).

Execute action (stub - real impl dispatches to handlers).

Execute agents respecting dependencies. (Simplified topological sort execution)

Core execute logic, called inside an optional tracing span.

Execute the determined recovery strategy. Args: strategy: Recovery strategy to execute trace_id: Trace ID

Execute a single workflow step. Returns (success, output) tuple.

Execute with timeout enforcement

Decrement depth; restore ``sys.modules`` on the outermost exit.

Export audit results to a CSV file.

Export a single SRE span as an OTEL span. Args: sre_span: The Agent SRE span to export sre_trace: The parent trace (f

Extract agent DID from the X-Agent-DID header.

Extract agent DID from request headers.

Return the set of anchor slugs defined by headings in ``markdown``.

Extract the raw callable from a LangGraph node spec. LangGraph 1.x stores nodes as ``StateNodeSpec(runnable=RunnableCallable)``. The

Extract claims from document

Extract code sections (functions, classes, etc.).

Extract common metadata patterns from episodes.

Extract searchable text content from state.

Extract context from episodes.

Extract fields from a dict or typed object.

Extract BOM fields from audit log entries.

Extract high-level intent from file change.