| name | operational-closure |
| description | Produce release-readiness, monitoring, rollback, and feedback artifacts that close the loop after implementation and verification - Brought to you by microsoft/hve-core |
Operational Closure
Turn "implemented" into "safely absorbed by the running system". This skill creates the artifacts and decisions that close the loop after code review, CI, and runtime verification.
When to Use
Invoke when a feature, fix, or risky change is ready to hand off into merge, deployment, or post-deploy monitoring.
Inputs
mode: (Required) One of pre-merge, post-merge, or post-deploy.
context: (Required) PR, task, feature, or release context.
verification_report: (Optional) Path to the runtime verification report.
Output
- Closure artifact at
docs/closure/{YYYY-MM-DD}-{slug}-closure.md
- Explicit go/no-go or ready/blocked recommendation
- Follow-up tasks or compound-learnings triggers when needed
Required Protocol
When the agent-intercom capability pack is installed, follow
.github/instructions/agent-intercom.instructions.md: broadcast closure readiness, blocked states,
monitoring handoff details, and any rollback trigger that the operator should be aware of during
the validation window.
When the browser-verification capability pack is installed, also follow
.github/instructions/browser-verification.instructions.md: carry browser
verification evidence, human-checkpoint outcomes, and browser-specific post-deploy
checks into the closure artifact instead of treating them as side notes.
When the strict-safety capability pack is installed, also follow
.github/instructions/strict-safety.instructions.md: carry the risky
ProposedAction entries, their ActionRisk, approval path, and final
ActionResult into closure artifacts when those details matter to rollout,
monitoring, or rollback.
When the release-observability capability pack is installed, also follow
.github/instructions/release-observability.instructions.md: integrate the
monitoring plan (SLIs, dashboards, alert thresholds), pre-deploy audit results,
post-deploy observation window (owner, duration), and rollback triggers into
the closure artifact as structured sections rather than ad hoc notes.
Step 1: Gather Closure Context
Collect:
- Summary of the change
- CI status and unresolved review items
- Runtime verification report (required when runtime surfaces were changed) — including verdict (PASS / PASS WITH FOLLOW-UP / FAIL / BLOCKED), evidence, and follow-up recommendations. If verification was BLOCKED, record the blocked status and the missing prerequisite as a closure condition.
- Any risky actions that required approval, rollback planning, or explicit containment
- Affected runtime surfaces
- Deployment or release path, if applicable
- Invariants that must remain true after release
- Data, migration, or rollout-sensitive behavior that could require extra monitoring
Step 2: Build the Closure Checklist
The closure artifact MUST include:
- Invariants to preserve — the behaviors or guarantees that cannot regress
- Pre-deploy audits — migrations, flags, config, access, or rollout prerequisites that must be checked before release
- Deployment or rollout path — merge-only, deploy, canary, phased rollout, maintenance window, or handoff path
- Post-deploy checks — the first concrete observations or smoke checks to run after release
- Risky action record — the
ProposedAction entries that materially affected rollout or rollback, their ActionRisk, and final ActionResult
- Source artifact cleanup — stash entries removed (ID + reason tag) and deliberation artifacts archived (ID + reason tag) during this closure; list source artifact IDs not found (already cleaned up or never created). Omit this section when no features in the shipment have
source_stash_id or source_deliberation_id populated.
- Healthy signals — what success should look like
- Failure signals — what indicates rollback or intervention is needed
- Monitoring plan — logs, dashboards, alerts, or smoke checks to watch
- Rollback trigger — the condition that should halt or reverse the rollout
- Rollback procedure — the actual rollback or mitigation action to take when the trigger fires
- Validation window — how long the change should be watched
- Owner — who is responsible for observing or acting
Step 3: Record Readiness Status
Return one of:
- READY — merge/deploy can proceed with the recorded monitoring plan
- READY WITH CONDITIONS — proceed only if named conditions are satisfied
- BLOCKED — missing verification, unclear rollback path, or unresolved runtime risk
Step 4: Feed Back into the Harness
When the closure process reveals durable knowledge:
- Invoke
compound for new runtime, deployment, or workflow learnings
- Update documentation if the release process or architecture knowledge changed
- Add tuning proposals if the harness lacked a needed safety mode, verification pattern, or reviewer
Why This Skill Exists
Operational closure is the compositional bridge from code production to safe absorption. It makes runtime verification actionable, keeps PRs honest about monitoring expectations, and turns release outcomes into future harness improvements.
Quality Criteria
- Closure artifacts include concrete release and monitoring signals, not generic advice
- Pre-deploy and post-deploy checks are explicit when the change has runtime or rollout risk
- Risky actions and their outcomes are visible when they affect release safety
- Rollback triggers and rollback procedures are explicit and actionable
- Ownership and validation windows are recorded
- Durable runtime learnings are fed back into the harness when appropriate
Model Routing
This skill operates at Tier 2 (Standard) — closure artifact assembly from defined inputs.
Generated by autoharness | Template: operational-closure/SKILL.md.tmpl