| name | govern |
| description | Govern AI spend without surveilling your team — the AI-spend governance loop. Set OKRs, harden the right context (tray→slate), bounded agents surface a finding ready for ratification, ratify a policy to the provenance trail, get a spend-ROI report that cites its sources. The leader governs usage; the team's private context is never read. |
| disable-model-invocation | true |
| allowed-tools | Bash(node *) |
| argument-hint | [optional: a governance question] |
/govern — the AI-spend governance loop
You are running one AI-spend governance pass for an engineering leader (Rhea, the
governor). The loop: sense → frame (OKRs) → harden (tray→slate) → read → refuse →
find → ratify → report. The point is governance without surveillance — the
leader controls which context hardens; the team's private streams (Slack, calendar
logs) stay in the tray unless explicitly approved.
Flow
1. Run the orchestrator
node skills/govern/run.js
Runs offline in fixture mode (no key needed); uses real bounded-agent reads when
ANTHROPIC_API_KEY or the claude CLI is present. Add --emit-cut to also write
the 11-govern visual-demo seam JSON, or --json for machine output.
2. Render the pass, in order
- Frame — the OKR baseline (e.g. 60% product / 40% security).
- Slate — what was hardened from the tray, and what was excluded as low-signal /
privacy-sensitive (name the excluded streams; this is the surveillance line).
- Read — the four-register read: Diligence (the spend split), Synthesis (vs the
OKR frame, under-allocation), Judgment (the verdict), Outreach (the route).
- Finding — lead with it. F-CAL (route verified calendar leak → CalendarOps) and
the dropped F-E14 (a false positive the adversarial cross-check removed via
PR-103). Show the reconciliation: naïve − dropped = verified realizable.
- Refuse — show the out-of-lane refusal verbatim (refusal is the routing).
- Ratify — the decision + the provenance hash (local-first chain).
- Report — the spend-ROI: verified savings/mo + /yr, and the source ids it cites.
3. Offer the next step
Point at the visual demo: open cuts/11-govern.html?run=govern-run.json (after
--emit-cut) to see the same pass render in the desktop surface. Offer to ratify a
different policy or rebalance the under-allocated objective.
Boundaries
- Never read or surface the team's private context (Slack/calendar) without approval.
- The finding is a governance decision ready for ratification, not a passive dashboard.
- The deterministic spine is the contract; live agent reads are additive.
