Enforce traceable, evidence-based findings with confidence labels and falsification checks.
npx skills add https://github.com/humblemuzzu/ghosttyyy --skill review-humblemuzzu
Please help me install this Agent Skill. Skill: review Source: https://github.com/humblemuzzu/ghosttyyy/tree/main/pi-setup/config-skills/review Command: npx skills add https://github.com/humblemuzzu/ghosttyyy --skill review-humblemuzzu
Epistemic standards for evaluation and analysis are critical to producing defensible findings. This skill provides a structured approach to load before reviewing code, debugging, reporting findings, or any task where claims must be defensible and traceable.
Load this skill before reviewing code, debugging, or reporting findings to ensure every claim is traceable and defensible.
High-intent search queries and answers about installing and using this skill.
To ensure code review findings are traceable, load this skill before evaluation to enforce a trace-or-delete rule, requiring every claim to link directly to code, logs, or data.
Labeling confidence during a debugging audit involves marking claims as VERIFIED, HUNCH, or QUESTION, which provides clear epistemic standards for analytical reporting and hypothesis tracking.
Applying falsification checks during analytical reporting requires designing tests specifically to disprove your hypotheses before confirming them, ensuring defensible and evidence-based findings.
Yes, you can use this evaluation skill for debugging tasks without dependencies, as it provides a structured reasoning approach for audits and analytical reporting requiring traceable evidence.
You need explicit traceability standards for evaluation tasks when claims must be defensible, such as during code reviews, debugging, or audits where evidence-based reasoning is critical.
Evidence-based code review without falsification testing lacks epistemic rigor; designing tests to disprove hypotheses before confirming them is required to produce defensible and traceable findings.
nexxelnEvaluate code findings with evidence-based criteria and confidence labeling.
14790897Validate analytical reports and SQL queries for methodological soundness and calculation accuracy.
HuaGCSVerify code, results, and configurations with structured evidence and auditable findings.
Chang-Shih-YungAggregate type checker, linter, and test results into a 0-10 composite score.
