BC save-usage
Auto-detect task complexity for OpenAI oAuth models (gpt-5.1-codex-mini + gpt-5.3-codex). Route only safe/negligible tasks to gpt-5.1-codex-mini. Use gpt-5.3-codex for anything executed, uncertain, or high-impact. RULES: Classify first. Default model: gpt-5.3-codex. Use gpt-5.1-codex-mini ONLY for safe triage/summarize/extract/reformat/dedupe/prompt-drafts/non-executable Q&A. NEVER use mini for security, auth, secrets, architecture, migrations, brownfield refactors, integration contracts, schema mapping, ordering/idempotency/retries, code changes, tool runs, or decisions that are hard to reverse. Reasoning: start LOW for mini. For 5.3-codex use LOW/MEDIUM by default; escalate to HIGH/EXTRA HIGH when 2+ are true: hard to reverse, affects 2+ domains (infra/data/security/ops/cost), subtle/expensive failure modes, long dependency-chain reasoning. ESCALATE immediately to gpt-5.3-codex if any: output will be executed (tools/code), ambiguity remains after 1 pass, contradictions found, or requirements are multi-constraint/structured output.
Auto-detect task complexity for OpenAI oAuth models (gpt-5.1-codex-mini + gpt-5.3-codex). Route only safe/negligible tasks to gpt-5.1-codex-mini. Use…
As a process C 55/100 · Has gaps — weak spots: result and completion, inputs and preconditions, progress reporting
How to improve
- Shorten the description to 1024 characters.
- 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
- error
description-longdescription is 1048 chars, limit 1024
Process rating: all ten parameters 55/100
- 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
- 50When it triggers. No condition that starts the skill
- 50Failures and branches. 0 branches, has a failure section
- 60Tools and files. Uses tools (bash) that frontmatter does not declare
- 100Steps. 29 steps
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 1054 tokens
- 100Running it twice. No mutating operations
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
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
- +3Description length 1048: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +4Description says when NOT to use the skill
- +4Structure: 10 headings
- +3Step-by-step instructions: 29 items
- +4Has examples (1 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 60.