BC santa-method
Multi-agent adversarial verification: two independent reviewers with the same rubric must both pass before output ships, with a fix-and-re-review convergence loop and human escalation cap. Use when gating publishing, production deploys, compliance or brand-sensitive content, or hallucination-prone claims before they ship.
Skill deploys two independent reviewers with matching rubric—both must sign off before content ships. If they disagree, it loops through fixes and re-reviews until consensus or human escalation. Promises a gate for publishing, production, compliance, and sensitive content.
Checks show no critical errors, quality at 84, process at 57. Single file, 2984 tokens. Review logic is there, but sandbox never ran and no model tests happened. In practice, you pay for two model calls per review cycle and must wire your own rubric—no template included.
Set it up if you're blocking critical content and can stomach double inference costs. Overkill for routine or low-stakes material.
Multi-agent adversarial verification: two independent reviewers with the same rubric must both pass before output ships, with a fix-and-re-review convergence…
As a process C 57/100 · Has gaps — weak spots: result and completion, inputs and preconditions, running it twice
How to improve
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 1. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 57/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 30Running it twice. 3 mutating operations with no state check
- 50When it triggers. No condition that starts the skill
- 50Failures and branches. 0 branches, has a failure section
- 85Steps. 38 steps, 1 vague phrases
- 100Tools and files. No external tools needed
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 2984 tokens
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +3Description length 323: enough signal without eating the budget
- +4Structure: 18 headings
- +3Step-by-step instructions: 38 items
- +4Has examples (10 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 84.