task
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It ingests data from an external MCP server and displays it to the user. (1) Ingestion points: Task titles and descriptions are retrieved from the witan server via task_ready() and task_list(). (2) Boundary markers: No specific delimiters are implemented to separate task content from agent instructions. (3) Capability inventory: Tool access is limited to task state transitions such as task_claim, task_create, and task_close. (4) Sanitization: The instructions do not define sanitization for retrieved content. This vulnerability surface is inherent to the skill's primary purpose and is considered a low risk.
- [DATA_EXFILTRATION]: The skill reads the local .git/config file to identify the repository's canonical HTTPS URI. This behavior is standard for development-oriented tools and does not involve the exfiltration of sensitive credentials or unauthorized network communication.
Audit Metadata