The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes data from external training logs and tracker files which presents an indirect prompt injection surface.
Ingestion points: Reads data from data/rehabilitation-tracker.json and daily logs in data/rehabilitation-logs/.
Boundary markers: There are no explicit instructions or delimiters used to prevent the agent from following commands that might be embedded in the health data.
Capability inventory: The skill has permissions for Read, Grep, Glob, Write, and Edit, which could be exploited if malicious instructions were successfully injected through data files.
Sanitization: No sanitization or validation logic is specified for the text content within the JSON data sources.
[COMMAND_EXECUTION]: The skill utilizes several file-system and text-processing tools to perform its analysis.
Evidence: The skill is configured to use Read, Grep, Glob, Write, and Edit tools to access and manage the rehabilitation data stored in the local file system.

rehabilitation-analyzer