Developer preview — docs versioned from day one.
Recipe library

Copy-paste mission specs

Six starter recipes for common governed business missions, written against the v1 draft mission-spec format. Adapt the tenant, identity, and policy fields to your own deployment.

Illustrative — not executable

Every recipe on this page is an illustrative example of the mission-spec shape. There is no public submission endpoint yet; field names, policy rule sets, and action kinds may change as the platform ships. Adapt freely, verify against the current docs before relying on anything.

1. Stale quote follow-up

salesillustrative

Find open quotes older than 7 days, draft a personalized follow-up for each, hold every draft for human approval, and log the touch in the CRM.

{
  "name": "stale-quote-followup",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "sales", "environment": "production" },
  "contract": {
    "goal": "Every open quote older than 7 days gets a follow-up touch",
    "definition_of_done": "CRM shows a follow-up touch per qualifying quote",
    "max_cost_usd": 25,
    "risk_bounds": ["no external sends without human approval"]
  },
  "identity": { "type": "service", "id": "missions/sales-followup" },
  "policy": { "rule_set": "sales-standard-v1" },
  "actions": [
    { "kind": "read", "target": "quotes.open_older_than_7d" },
    { "kind": "draft", "target": "followup_message", "requires_approval": true },
    { "kind": "record", "target": "crm.followup_touch" }
  ],
  "evidence": { "record": ["quotes_touched", "drafts_held_for_approval", "actual_cost_usd"], "compare": "expected_vs_actual" }
}

2. Daily invoice reconciliation

financeillustrative

Pull yesterday's invoices and payments, match them, and surface the unmatched remainder as an exception report — read-only until a human decides.

{
  "name": "daily-invoice-reconciliation",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "finance", "environment": "production" },
  "contract": {
    "goal": "All of yesterday's invoices matched to payments, exceptions listed",
    "definition_of_done": "Exception report delivered; zero silent unmatched items",
    "max_cost_usd": 10,
    "risk_bounds": ["read-only: no ledger writes"]
  },
  "identity": { "type": "service", "id": "missions/finance-recon" },
  "policy": { "rule_set": "finance-readonly-v1" },
  "actions": [
    { "kind": "read", "target": "billing.invoices_yesterday" },
    { "kind": "read", "target": "billing.payments_yesterday" },
    { "kind": "compute", "target": "match_invoices_to_payments" },
    { "kind": "report", "target": "exceptions.unmatched", "requires_approval": false }
  ],
  "evidence": { "record": ["invoices_seen", "matched_count", "exceptions_count", "actual_cost_usd"], "compare": "expected_vs_actual" }
}

3. Deployment health check

engineeringillustrative

Every 15 minutes, check production endpoints and certificate status; page only on two consecutive failures so transient blips don't wake anyone.

{
  "name": "deployment-health-check",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "engineering", "environment": "production" },
  "contract": {
    "goal": "Detect real outages within 15 minutes, ignore transient blips",
    "definition_of_done": "Alert fires only after 2 consecutive failures",
    "max_cost_usd": 5,
    "risk_bounds": ["no deploys, no config changes"]
  },
  "identity": { "type": "service", "id": "missions/health-check" },
  "policy": { "rule_set": "engineering-monitor-v1" },
  "actions": [
    { "kind": "probe", "target": "endpoints.production" },
    { "kind": "probe", "target": "tls.cert_expiry" },
    { "kind": "alert", "target": "oncall", "condition": "two_consecutive_failures" }
  ],
  "evidence": { "record": ["probes_run", "failures_seen", "alerts_fired", "actual_cost_usd"], "compare": "expected_vs_actual" }
}

4. Morning approval-inbox digest

operationsillustrative

Each morning, collect everything waiting in the approval inbox, rank it by contract risk, and deliver one digest — so approvals are a decision, not a scavenger hunt.

{
  "name": "morning-approval-digest",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "operations", "environment": "production" },
  "contract": {
    "goal": "One ranked digest of pending approvals by 8am",
    "definition_of_done": "Digest delivered; every pending item ranked by risk",
    "max_cost_usd": 8,
    "risk_bounds": ["digest only: never approves on anyone's behalf"]
  },
  "identity": { "type": "service", "id": "missions/ops-digest" },
  "policy": { "rule_set": "operations-readonly-v1" },
  "actions": [
    { "kind": "read", "target": "approvals.pending" },
    { "kind": "compute", "target": "rank_by_contract_risk" },
    { "kind": "report", "target": "digest.morning", "requires_approval": false }
  ],
  "evidence": { "record": ["items_pending", "digest_delivered_at", "actual_cost_usd"], "compare": "expected_vs_actual" }
}

5. Contract renewal reminder

operationsillustrative

Scan vendor and customer contracts expiring in the next 60 days, draft renewal briefs with the key terms, and hold each brief for review before anyone acts.

{
  "name": "contract-renewal-reminder",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "operations", "environment": "production" },
  "contract": {
    "goal": "No contract auto-renews or lapses unnoticed",
    "definition_of_done": "Renewal brief exists for every contract expiring within 60 days",
    "max_cost_usd": 12,
    "risk_bounds": ["no signatures, no terms accepted, drafts only"]
  },
  "identity": { "type": "service", "id": "missions/contract-watch" },
  "policy": { "rule_set": "operations-standard-v1" },
  "actions": [
    { "kind": "read", "target": "contracts.expiring_within_60d" },
    { "kind": "draft", "target": "renewal_brief", "requires_approval": true },
    { "kind": "record", "target": "contracts.renewal_brief_logged" }
  ],
  "evidence": { "record": ["contracts_seen", "briefs_drafted", "actual_cost_usd"], "compare": "expected_vs_actual" }
}

6. Weekly business-metrics report

reportingillustrative

Every Monday, assemble the week's numbers — revenue, cost, active missions — compare them against their contracts, and publish the report to the team's channel.

{
  "name": "weekly-metrics-report",
  "version": "1.0-draft",
  "tenant": { "organization": "example-org", "workspace": "leadership", "environment": "production" },
  "contract": {
    "goal": "Weekly numbers published every Monday by 9am",
    "definition_of_done": "Report published; expected-vs-actual included per metric",
    "max_cost_usd": 10,
    "risk_bounds": ["read-only sources; publishes to one configured channel"]
  },
  "identity": { "type": "service", "id": "missions/weekly-report" },
  "policy": { "rule_set": "reporting-standard-v1" },
  "actions": [
    { "kind": "read", "target": "metrics.revenue_week" },
    { "kind": "read", "target": "metrics.cost_week" },
    { "kind": "read", "target": "missions.completed_week" },
    { "kind": "report", "target": "channel.weekly_metrics", "requires_approval": false }
  ],
  "evidence": { "record": ["metrics_included", "published_at", "actual_cost_usd"], "compare": "expected_vs_actual" }
}
Using these recipes

Start from the recipe closest to your mission, then change three things: the tenant (yours, exactly one), the identity (your verified service identity), and the policy.rule_set (the deterministic rule set your deployment actually enforces). The contract's risk bounds are the part that keeps the mission governed — tighten them before you loosen anything else.