Execution Evidence Gate

Gate ID: execution-evidence

Enforces that an artifact has signed execution evidence from a specific environment before promotion. Ensures artifacts are observed running (with sufficient trace quality) before advancing through the release pipeline.

How It Works

  1. Checks if the target environment requires execution evidence (configurable per environment)
  2. Reads execution evidence metadata from the policy context
  3. If no evidence exists, applies the configured action (warn or block)
  4. If evidence exists, validates trace quality (minimum hot symbols and unique call paths)
  5. Returns pass, warn, or block result

Configuration

{
  "PolicyGates": {
    "ExecutionEvidence": {
      "Enabled": false,
      "MissingEvidenceAction": "Warn",
      "RequiredEnvironments": ["production"],
      "MinHotSymbolCount": 3,
      "MinUniqueCallPaths": 1
    }
  }
}

Options

OptionTypeDefaultDescription
EnabledboolfalseWhether the gate is active (opt-in)
MissingEvidenceActionenumWarnAction when evidence is missing: Warn or Block
RequiredEnvironmentsstring[]["production"]Environments that require execution evidence
MinHotSymbolCountint3Minimum hot symbols for sufficient trace quality
MinUniqueCallPathsint1Minimum unique call paths for sufficient trace quality

Context Metadata Keys

The gate reads the following keys from PolicyGateContext.Metadata:

KeyTypeDescription
has_execution_evidence"true"/"false"Whether execution evidence exists
execution_evidence_hot_symbol_countint stringNumber of hot symbols in the evidence
execution_evidence_unique_call_pathsint stringNumber of unique call paths

Example Gate Results

Pass (evidence meets quality):

Execution evidence meets quality thresholds (hot symbols: 42, call paths: 17)

Pass (environment not required):

Execution evidence not required for environment 'staging'

Warn (no evidence, warn mode):

No execution evidence found for this artifact (warn mode)

Fail (no evidence, block mode):

No execution evidence found for this artifact in required environment

Fail (insufficient quality):

Execution evidence trace quality is insufficient: hot symbols 1 < 3 or call paths 0 < 1

Integration

This gate consumes stella.ops/executionEvidence@v1 predicates generated by the Signals execution evidence pipeline. Evidence is populated in the policy context during release evaluation.

Typical flow:

  1. Artifact runs in staging environment
  2. Signals captures runtime trace via eBPF/ETW
  3. ExecutionEvidenceBuilder generates signed predicate
  4. Release promotion to production triggers policy evaluation
  5. This gate verifies execution evidence exists from staging

Last updated: 2026-02-19.