The Agent Skills Directory

[SAFE]: No security issues detected. The skill is focused on game testing and regression diagnosis.
[COMMAND_EXECUTION]: The skill provides a Python script, scripts/trace_compare.py, for comparing game trace files. Analysis of the script confirms it uses only the Python standard library and includes several security-conscious design choices:
Strict Validation: The script validates the JSON schema, data types, and nesting depth of input files.
Resource Limits: It enforces a 16 MiB size limit for input files to prevent memory-related issues.
Safe File Writing: It uses the 'x' open mode for the output report, which ensures that existing files are never overwritten.
No Risky Operations: The script does not perform network requests, execute shell commands, or use dynamic execution functions like eval or exec.
[DATA_EXPOSURE]: There is no evidence of the skill attempting to access sensitive system files, environment variables, or credentials. It operates on structured game data (JSON/JSONL).
[INDIRECT_PROMPT_INJECTION]: The skill ingests external data in the form of gameplay tapes and traces. This ingestion surface is mitigated by strict schema validation in the accompanying script, and the content is treated as data for comparison rather than as instructions for the agent.

gameplay-harness