The Agent Skills Directory

[SAFE]: Extensive analysis of the skill's code and logic revealed no malicious patterns or security vulnerabilities. The tool follows established security practices for local static analysis tools.
[COMMAND_EXECUTION]: The script scripts/audit_paper.py invokes internal analysis scripts using subprocess.run. This process is securely implemented using explicit paths and the current Python executable, avoiding shell injection risks.
[REMOTE_CODE_EXECUTION]: There are no network-based operations or external downloads. All executable logic is contained within the skill's distribution, precluding the risk of remote code injection.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes user-provided manuscripts.
Ingestion points: Local file reading is performed in scripts/check_tex.py and scripts/check_bib.py.
Boundary markers: No specific delimiters are used to isolate ingested text from instructions during file processing.
Capability inventory: Capabilities are limited to local file system reads and execution of bundled Python scripts.
Sanitization: The skill employs regex-based heuristic analysis; while it does not perform sanitization of text ingested into the LLM context, the risk is mitigated by the tool's limited local scope.

stat-writing