Audit root cause traceback evidence integrity

What it's for

Makes the site's anomaly-to-root-cause traceback claim precise: Gitrevio can distinguish a temporal precursor from a causally identified path and abstain when the evidence does not support the stronger wording.

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": 150,
  "candidate_screenings": [
    {
      "adjusted_p_value": 0.02,
      "candidate_metric_id": "review-delay",
      "changepoint_at_ms": 50,
      "eligible": true,
      "evidence_verified": true,
      "id": "screen",
      "observed_through_ms": 100,
      "raw_p_value": 0.01,
      "screened": true,
      "source_artifact_id": "artifact",
      "traceback_case_id": "case"
    }
  ],
  "dependency_edges": [
    {
      "edge_basis_id": "quasi_experimental",
      "effect_estimate": 0.3,
      "effect_interval_lower": 0.1,
      "effect_interval_upper": 0.5,
      "evidence_verified": true,
      "graph_version_id": "graph-v1",
      "id": "edge",
      "maximum_lag_ms": 100,
      "minimum_lag_ms": 10,
      "observed_through_ms": 20,
      "source_artifact_id": "artifact",
      "source_metric_id": "review-delay",
      "support_count": 100,
      "target_metric_id": "delivery-delay"
    }
  ],
  "evidence_artifacts": [
    {
      "id": "artifact",
      "immutable_hash": "abc",
      "independently_verified": true,
      "observed_at_ms": 20
    }
  ],
  "graph_versions": [

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.

{
  "case_diagnostics": [
    {
      "claim_type_id": "causal_root",
      "decision": "traceback_supported",
      "eligible_ancestor_count": 1,
      "failed_gates": [],
      "passing_validation_slice_count": 1,
      "screened_ancestor_count": 1,
      "selected_root_diagnostics": [
        {
          "failed_gates": [],
          "metric_id": "review-delay",
          "path_edge_basis_ids": [
            "quasi_experimental"
          ],
          "path_edge_count": 1
        }
      ],
      "selected_root_metric_ids": [
        "review-delay"
      ],
      "target_metric_id": "delivery-delay",
      "traceback_case_id": "case"
    }
  ],
  "configuration": {
    "as_of_ms": 150,
    "maximum_false_attribution_rate": 0.2,
    "minimum_edge_support": 30,
    "minimum_top_k_recall": 0.7,
    "minimum_validation_cases": 30,
    "require_causal_edges_for_causal_claim": true,
    "significance_level": 0.05
  },
  "decision": "use_tracebacks",
  "failed_gates": [],
  "guardrails": [
    "Temporal precedence and graph depth do not establish causation. A causal-root label requires every selected path edge to have randomized, quasi-experimental or independently justified mechanistic identification.",
    "Every upstream candidate must be represented on a zero-inclusive screening lattice and multiplicity-adjusted; reporting only the winning path creates selection bias.",
    "Results are aggregate diagnostic evidence, not authority to blame, discipline or make employment decisions about a person. Human review and a correction route remain mandatory."
  ],
  "method": "point_in_time_root_cause_traceback_evidence_audit_v1",
  "summary": {

How it works

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

1. 1 Freeze the graph version effective at each incident and reconstruct its acyclic metric graph from immutable evidence available at the case cutoff.
2. 2 Require a zero-inclusive screen of every upstream candidate, adjusted significance, an in-window earlier changepoint and supported non-null edges on every selected path.
3. 3 Permit causal-root language only when every path edge is randomized, quasi-experimental or independently mechanistic, and require strictly later validation of root recall and false attribution.

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 whether an anomaly traceback is" }
  → finds "audit_root_cause_traceback_evidence_integrity"

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

gitrevio_capability_run
  { "capability_id": "audit_root_cause_traceback_evidence_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 Quality, incidents & reliability →