Optimize causal release assurance portfolio

What it's for

Converts a release-risk score into the economically valuable next question: which review, test, canary, staged rollout or hold action has demonstrated incremental protection and deserves scarce release capacity?

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.

{
  "assurance_options": [
    {
      "change_id": "payments-release",
      "cost": 0,
      "delay_cost_per_hour": 0,
      "delay_hours": 0,
      "evidence_verified": true,
      "id": "standard-release",
      "mode_id": "standard",
      "provided_control_ids": [
        "peer-review"
      ],
      "releases_change": true,
      "resource_requirements": {
        "release-engineer-hours": 0
      }
    },
    {
      "change_id": "payments-release",
      "cost": 2500,
      "delay_cost_per_hour": 500,
      "delay_hours": 2,
      "evidence_verified": true,
      "id": "canary-release",
      "mode_id": "canary",
      "provided_control_ids": [
        "peer-review",
        "canary"
      ],
      "releases_change": true,
      "resource_requirements": {
        "release-engineer-hours": 2
      }
    }
  ],
  "baseline_mode_id": "standard",
  "budget": 5000,
  "changes": [
    {
      "base_failure_probability": 0.3,
      "change_class_id": "service-change",
      "evidence_verified": true,
      "failure_loss": 250000,

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": [
    "Accepted randomized or quasi-experimental studies identify the assurance mode's effect for the submitted change class after the declared transport weight; consistency, positivity and checked interference remain required.",
    "Failure losses, release values, common-risk groups, resource capacities and mandatory controls are organization-owned decision inputs rather than values learned from another tenant."
  ],
  "decision": "release_assurance_portfolio_selected",
  "evidence": {
    "accepted_effects": [
      {
        "accepted_studies": 1,
        "change_class_id": "service-change",
        "eligible": true,
        "median_relative_risk": 0.283,
        "mode_id": "canary",
        "probability_of_benefit": 1,
        "relative_risk_p95": 0.4903
      }
    ],
    "rejected_options": [],
    "rejected_studies": 0
  },
  "limitations": [
    "Relative-risk posteriors do not make observational risk factors causal; unmeasured study confounding, implementation drift and changing release systems can invalidate transport.",
    "The result is a human-governed assurance recommendation. It does not autonomously merge or block a change and never treats an author, reviewer or team as the cause of release risk."
  ],
  "method": "causal_beta_binomial_common_risk_assurance_portfolio_v1",
  "pareto_frontier": [
    {
      "cost": 0,
      "cvar_loss": 252083.3333,
      "expected_loss": 73600,
      "expected_net_value": 106400,
      "option_ids": [
        "standard-release"
      ]
    },
    {
      "cost": 2500,
      "cvar_loss": 253500,
      "expected_loss": 22600,
      "expected_net_value": 157400,
      "option_ids": [
        "canary-release"
      ]

How it works

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

1. 1 Validate randomized or propensity-known preregistered quasi-experimental assurance studies, discount them by declared transport weight and estimate class-by-mode posterior relative-risk distributions.
2. 2 Reject modes without mature interference-checked evidence or required controls, then simulate every feasible option portfolio with common random numbers and count each shared risk-group event once.
3. 3 Choose the highest risk-adjusted net-value portfolio that clears budget, capacity, expected-failure and CVaR gates; disclose exact versus beam search and the cost/loss/value Pareto frontier.

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": "select one release assurance or hold" }
  → finds "optimize_causal_release_assurance_portfolio"

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

gitrevio_capability_run
  { "capability_id": "optimize_causal_release_assurance_portfolio", "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 Delivery forecasting & commitments →