The Agent Skills Directory

[SAFE]: The skill operates as a high-level router and aggregator for mathematical verification. It defines clear procedures for decomposing theorems into atomic claims and dispatching them to appropriate computational or formal verification tools (R0-R3).
[SAFE]: The provided Python script scripts/extract_block.py is a utility for parsing LaTeX files using standard library regular expressions. It identifies theorem environments and referenced equations to create self-contained blocks for verification. The script performs no network operations, executes no external commands beyond standard file I/O, and contains no obfuscated logic.
[SAFE]: Tool usage (allowed-tools) is appropriate for the stated purpose of reading source files, extracting content, and writing verification reports. The inclusion of Bash and Skill tools is consistent with the need to invoke the extraction script and other verification rungs.
[SAFE]: The skill follows secure patterns for handling untrusted content by advising the use of a fresh-context agent (domain-reviewer) for conceptual reviews, ensuring that the review process is isolated from the context that generated the math.

verify-math