metabolights-skill
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [File System Interaction Capability]: The
scripts/rest_request.pyscript includes functionality to save API responses to a local file path specified by theraw_output_pathparameter. This grants the skill the capability to create directories and write files to any location the agent's environment has permission to access, which is a potential consideration for maintaining system integrity and preventing unauthorized file modifications. - [Flexible Network Communication]: The URL construction logic in
scripts/rest_request.pyallows thepathparameter to override the defaultbase_urlif a full URL is provided. This allows the skill to communicate with arbitrary external services beyond the intended MetaboLights domain, which could be used to facilitate data transfer to non-primary endpoints. - [Indirect Prompt Injection Surface]: The skill processes data from external APIs and returns summaries to the agent, creating a surface where malicious instructions embedded in the retrieved data could influence behavior.
- Ingestion points: Data enters the agent context through API responses fetched in
scripts/rest_request.py. - Boundary markers: None identified; the skill does not use explicit delimiters or warnings to separate external data from its own instructions.
- Capability inventory: The skill has capabilities for network communication (
requests) and file system writes (pathlib.Path.write_text) inscripts/rest_request.py. - Sanitization: External content is truncated for length via a
_compactfunction but does not undergo content-based sanitization or instruction filtering.
Audit Metadata