Confidence fragility review
Skill markoblogo/abvx-agent-skills/skills/confidence-fragility-review
Reviewable capability layer for coding agents. Portable skills, delivery gates, workflow patterns, and verification-first engineering.
npx -y skills add markoblogo/abvx-agent-skills --skill confidence-fragility-reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Review whether an artifact's confident claims are backed by evidence, tests, contracts, or operational safeguards. Use before shipping, publishing docs, accepting generated plans, or trusting a workflow that may present fragile assumptions as certainty.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
2.0 KB, as published. Nobody here has run it
Confidence Fragility Review
Stress-test the gap between how confident an artifact sounds and how much evidence actually supports it.
Use When
- a plan, release note, README, landing page, or SET bundle sounds ready but proof is thin;
- generated docs imply stronger guarantees than the repo can support;
- final review needs a fear-oriented lens before ship;
- a user asks what could go wrong if people trust the artifact at face value.
Workflow
- Identify confident claims:
- readiness claims;
- safety claims;
- completeness claims;
- compatibility claims;
- performance or reliability claims.
- For each claim, ask what backs it:
- tests or CI;
- schema or type contracts;
- runtime checks;
- docs links;
- human approval;
- explicit limitations.
- Classify the concern:
- tactical: a specific path can fail;
- structural: a category of failure has no defense;
- epistemic: confidence is not earned by evidence.
- Separate justified fragility from projected worry.
- Recommend the smallest evidence step that would make confidence warranted.
Output Shape
Use:
- verdict:
CONFIDENCE_WARRANTEDorFRAGILITY_MASKED; - confidence claims reviewed;
- evidence found;
- fragility register: tactical, structural, or epistemic;
- smallest proof step.
Guardrails
- Do not turn this into a generic risk audit.
- Do not demand proof for every harmless claim.
- Do not perform security analysis unless the artifact makes security claims.
- Be explicit when a fear is plausible but not evidenced.
Provenance
Adapted from the anxiety-reader pattern in aself101/agents-and-pipelines, reframed as a compact confidence-vs-evidence review skill.