Audit AI configuration release integrity

What it's for

Makes AI change management auditable: leaders can prove that production received the exact prompt/model/tool-policy bundle tested, on the declared applications, with live controls and a real rollback path.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "configuration_versions": [
    {
      "changed_component_ids": [],
      "configuration_hash": "sha256:support-v1",
      "created_at_ms": 1000,
      "evidence_verified": true,
      "id": "support-v1",
      "parent_version_id": null,
      "required_control_ids": [
        "rollback"
      ],
      "system_id": "support-agent",
      "version_sequence": 1
    },
    {
      "changed_component_ids": [
        "prompt",
        "tool-policy"
      ],
      "configuration_hash": "sha256:support-v2",
      "created_at_ms": 2000,
      "evidence_verified": true,
      "id": "support-v2",
      "parent_version_id": "support-v1",
      "required_control_ids": [
        "rollback"
      ],
      "system_id": "support-agent",
      "version_sequence": 2
    }
  ],
  "release_candidates": [
    {
      "approval_control_ids": [
        "rollback"
      ],
      "configuration_version_id": "support-v2",
      "declared_application_ids": [
        "support"
      ],
      "evaluation_completed_at_ms": 3000,
      "evaluation_passed": true,
      "evidence_verified": true,

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "assumptions": [
    "A configuration hash covers the exact prompt, model route, tool policy, retrieval policy, controls and runtime parameters evaluated and deployed; version and rollout records are immutable point-in-time facts.",
    "Declared applications form the prospective blast-radius perimeter, evaluation completes before exposure, required approvals remain effective at runtime and rollback points to a tested earlier configuration of the same system.",
    "Passing proves release lineage and control integrity, not that an evaluation is representative, a configuration is safe, or a rollout caused later business outcomes.",
    "The audit is system-level assurance, not employee monitoring, authorship inference, vendor scoring, content inspection or deployment authorization."
  ],
  "candidate_diagnostics": [],
  "configuration": {
    "minimum_evidence_coverage": 0.95
  },
  "decision": "ai_configuration_release_integrity_supported",
  "failed_gates": [],
  "method": "ai_configuration_evaluation_release_rollout_lineage_audit_v1",
  "summary": {
    "configuration_version_count": 2,
    "deployed_candidate_count": 1,
    "evidence_coverage": 1,
    "release_candidate_count": 1,
    "rollout_event_count": 2,
    "system_count": 1,
    "valid_candidate_count": 1
  },
  "truncated_candidate_count": 0,
  "truncated_version_count": 0,
  "version_diagnostics": []
}

How it works

Causal inference & experiment design — Separate what a change caused from what merely moved alongside it.

1. 1 Validate consecutive per-system configuration versions, exact parent ancestry, increasing creation time, unique full-bundle hashes and explicit changed components.
2. 2 Join each candidate to its evaluated configuration, approvals, declared applications and earlier same-system rollback version; deployment exposure requires a passing pre-rollout evaluation.
3. 3 Audit consecutive non-overlapping stages, runtime controls, observed-versus-declared blast radius, monotone traffic or closed rehearsed rollback, target reconciliation and evidence coverage.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "audit that the exact immutable ai" }
  → finds "audit_ai_configuration_release_integrity"

gitrevio_capability_describe
  { "capability_id": "audit_ai_configuration_release_integrity" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "audit_ai_configuration_release_integrity", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in AI risk, rights & assurance →