Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →boshu2/anti-ceremony
Guard outcome work against process overhead. Triggers: RPI pre-Plan guard; explicit "full anti-ceremony audit" requests.
Passed 18 of 23 test cases with the skill, and 6 of 23 without it.
and 13 more in the full report
13 fixed5 passed either way4 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
Perform a quick guard assessment on our database upgrade plan. Outcome: Upgrade payment microservice database to PostgreSQL 15. Proposed process work: Create a secondary staging metrics dashboard and write a ten-page architecture summary document. Remaining proof: Migration script executes cleanly on staging DB backup with zero data loss validation. Stop condition: Production database upgraded to v15 and health checks green. Format your response exclusively as YAML containing decision (CONTINUE or STOP), reason (one sentence), frozen_outcome, parked_process_work, remaining_proof, and stop_condition.
| Checked | Without skill | With skill |
|---|---|---|
| The parked_process_work field is structured as a YAML list. | Missed | Passed |
| The decision field value is set to CONTINUE. | Passed | Passed |
The instructions your model reads when the skill loads.
Keep work tied to the caller-visible outcome and the proof still needed for it. The optional guard is a single, artifact-free judgment when selected; the full honesty audit runs only on an explicit "full anti-ceremony audit" request.
Quick guard. Outcome: "ao gate check lists the probe-coverage row". Proposed
process work: a coverage dashboard and a second audit pass. Remaining proof:
the gate row, green in CI. Stop condition: that gate passes on main.
.agents/, and no tracker write is issued.reason: is one sentence, and parked_process_work: names the
dashboard-shaped items.full anti-ceremony audit.STOP when the traversal would only create control artifacts or the
frozen outcome is already sufficiently proved; otherwise return CONTINUE.Progress needs evidence tied to acceptance or a named blocking uncertainty.
Digest movement, finding counts, and additional reviews alone earn no credit.
Informative red can justify a different experiment inside an explicitly selected
authorized outcome's unchanged acceptance and allowance; repeated no-information work
cannot. Keep required independent proof in remaining_proof, including every
necessary unresolved finding. Apply Validate's effect-based risk rule to the
cost of proof; never weaken the exact-subject or all-acceptance PASS bar.
test cases passed, vs 6 / 23 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.