What problem does it solve?
Convert academic papers and research articles into validated, actionable improvements for an agent skill library by extracting empirical findings, assessing credibility, running security checks, and mapping recommendations to existing skills.
Core Features & Use Cases
- Credibility assessment: Apply a 6-dimension rubric to score papers and gate extraction, with explicit rules for borderline and reject cases.
- Security-first extraction: Invoke secure-* security checks and block any content that fails the security pipeline before taking action.
- Insight extraction and application: Classify findings (TECHNIQUE, GOTCHA, FAILURE_MODE, METRIC), recommend APPLY/PARTIAL/KEEP CURRENT, map insights to skills, and optionally apply changes with validation and citation logging.
- Use Case: Feed an uploaded PDF or arXiv link to produce a credibility report, extracted insights with recommendations, and a prioritized application plan for improving relevant skills.
Quick Start
Analyze the uploaded paper PDF and return a credibility score, security verdicts, extracted insights with recommendations, and an application plan with citations.