The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to execute local Python scripts (process_analyzer.py, handoff_analyzer.py, and improvement_scorer.py) via the command line to analyze process data.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze process data from external JSON files (e.g., assets/sample_process.json), creating a vulnerability surface where malicious instructions could be embedded in the input data to manipulate the agent's actions. (1) Ingestion points: Local JSON input files processed by the scripts in the scripts/ directory. (2) Boundary markers: None identified in the instructions or script logic to isolate untrusted data. (3) Capability inventory: Subprocess execution of Python scripts and local file reading. (4) Sanitization: The scripts use standard JSON parsing, which validates structure but does not sanitize content against potential natural language injection attacks.

process-mapper