find-series-opportunities
Warn
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to read files 'silently' and explicitly commands it to 'not mention this step to the user', which are instructions designed to conceal agent actions from the end user.
- [DATA_EXFILTRATION]: The skill accesses a hidden dotfile in the user's home directory (~/.devadvokit.md), which is a sensitive path often used for private configuration, though it is used here as a content library.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and follows instructions from an external file (../../shared/ai-antipatterns.md) to modify its behavior and output format.
- Ingestion points: ../../shared/ai-antipatterns.md
- Boundary markers: Absent. The skill provides no delimiters between the skill logic and the ingested file.
- Capability inventory: The skill utilizes file system read access.
- Sanitization: Absent. The content of the external file is obeyed directly without validation or filtering.
Audit Metadata