The Agent Skills Directory

[SAFE]: The skill implements a robust framework for patent analysis (FTO, infringement, invalidation) using strict evidence-status gates (A-Confirmed to D-Inconclusive) to prevent the AI from generating hallucinated certainties.
[SAFE]: Includes a suite of Python scripts (check_legal_sources.py, validate_skill.py, check_evals.py) intended for maintainers to perform static analysis on the skill's legal source data and evaluation assertions. These scripts use standard libraries and contain no malicious execution patterns.
[SAFE]: The skill's configuration and evaluation files contain defensive patterns (regex) specifically designed to block the AI from outputting external URLs or potentially hallucinated legal success rates.
[SAFE]: No hardcoded credentials, sensitive file access, or unauthorized network operations were found in the instructions or the provided scripts.
[SAFE]: The skill's instructions emphasize the role of the AI as a preliminary analyzer and explicitly state that all outputs must be reviewed by professional patent lawyers/agents.

patent-analysis