The Agent Skills Directory

[Generic Network Interaction]: The script scripts/rest_request.py acts as a flexible REST client. While the skill instructions direct the agent to use a specific public API, the script's design allows for requests to any URL provided in the input. This is a common pattern for utility scripts, but it relies on the agent's adherence to its operating rules to prevent unintended network operations.
[File System Interaction]: The skill includes functionality to save API responses to the local filesystem using the save_raw and raw_output_path parameters. This capability allows the agent to write data to arbitrary paths. While intended for saving search results, it warrants review to ensure that file operations are restricted to appropriate directories.
[Indirect Prompt Injection Surface]: This skill processes data from an external scientific database and possesses network and file-writing capabilities. This configuration represents a potential surface for indirect prompt injection, where instructions embedded in external data could influence subsequent agent behavior.
Ingestion points: External data is retrieved from API endpoints (e.g., www.ebi.ac.uk) via scripts/rest_request.py.
Boundary markers: The script does not implement specific delimiters or warnings to isolate external content from the agent's internal instructions.
Capability inventory: The skill uses subprocess execution (running the python script), network requests (via the requests library), and file system writes (path.write_text).
Sanitization: Input fields are validated for type and presence, but the content retrieved from external APIs is not sanitized before being returned to the agent or saved to a file.

chebi-skill