Audit AI capability fallback integrity

What it's for

Shows leaders which AI-dependent workflows can genuinely continue when the assistant, agent or provider disappears—not which runbooks merely exist on paper.

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.

{
  "as_of_ms": 10000,
  "capabilities": [
    {
      "approved_runbook_hash": "sha256:runbook-v2",
      "criticality_weight": 1,
      "domain_id": "platform",
      "evidence_verified": true,
      "id": "deployment-recovery",
      "maximum_activation_time_hours": 2,
      "maximum_evidence_age_days": 90,
      "required_fallback_units": 3
    }
  ],
  "fallback_exercises": [
    {
      "activation_time_hours": 1,
      "ai_disabled": true,
      "attempted_units": 4,
      "capability_id": "deployment-recovery",
      "evidence_verified": true,
      "exercise_group_id": "quarterly-2026-q3",
      "exercised_at_ms": 9000,
      "id": "fallback-exercise-2026-q3",
      "independent_observer": true,
      "runbook_hash": "sha256:runbook-v2",
      "successful_units": 4
    }
  ],
  "workflow_dependencies": [
    {
      "ai_provider_group_id": "agent-provider",
      "capability_id": "deployment-recovery",
      "dependency_fraction": 0.8,
      "evidence_verified": true,
      "fallback_required": true,
      "id": "release-dependency",
      "workflow_id": "agentic-release-workflow"
    }
  ]
}

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": [
    "Capabilities, workflows and exercises are aggregate operational units with stable IDs; exercise groups are not reconstructed from named-person activity.",
    "An exercise proves fallback only when AI is actually disabled, the approved runbook is current, sufficient units participate, activation is timely and independent evidence is retained.",
    "Dependency fractions describe workflow exposure and provider concentration, not employee contribution or intent."
  ],
  "capability_diagnostics": [
    {
      "capability_id": "deployment-recovery",
      "domain_id": "platform",
      "failed_gates": [],
      "fallback_required": true,
      "largest_provider_exposure_fraction": 1,
      "latest_activation_time_hours": 1,
      "latest_exercise_age_days": 0,
      "latest_exercise_id": "fallback-exercise-2026-q3",
      "latest_success_fraction": 1,
      "workflow_dependency_exposure": 0.8
    }
  ],
  "counts": {
    "capabilities": 1,
    "fallback_exercises": 1,
    "required_capabilities": 1,
    "verified_required_capabilities": 1,
    "workflow_dependencies": 1,
    "workflows": 1
  },
  "decision": "ai_fallback_capability_evidence_verified",
  "dependency_diagnostics": [],
  "evidence": {
    "gate_pass": true,
    "minimum_required_fraction": 0.95,
    "verified_fraction": 1
  },
  "integrity_pass": true,
  "limitations": [
    "A passing drill demonstrates represented conditions at one point in time; it does not guarantee performance in every outage, crisis or future workflow version.",
    "The audit cannot infer individual skill, effort, loyalty, misconduct, nationality, employment suitability or whether a person should be disciplined."
  ],
  "method": "aggregate_unaided_ai_fallback_evidence_reconciliation_v1",
  "truncated_detail_rows": 0
}

How it works

Statistical audit & measurement — Check whether a number is fit to decide on: coverage, timing, reconciliation, and the gaps a dashboard hides.

1. 1 Reconcile the capability registry to every aggregate AI-workflow dependency and preserve provider-group exposure rather than treating a workflow inventory as fallback proof.
2. 2 Select the latest non-future exercise per required capability and test evidence age, approved runbook hash, real AI disablement, attempted/successful unit sufficiency, activation time, independent observation and provenance.
3. 3 Report capability and dependency failures with evidence coverage and provider concentration while withholding all named-person inference.

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": "prove that every aggregate capability required" }
  → finds "audit_ai_capability_fallback_integrity"

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

gitrevio_capability_run
  { "capability_id": "audit_ai_capability_fallback_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 cost, routing & return →