backlog-workflow
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or security vulnerabilities were identified. The skill is limited to project management guidance and standard developer workflows.\n- [COMMAND_EXECUTION]: The skill incorporates instructions for performing git commits to link code changes to specific task IDs. This is a standard practice for development workflows and is consistent with the skill's primary purpose.\n- [PROMPT_INJECTION]: The skill uses procedural markers such as "CRITICAL" and "ALWAYS" to emphasize workflow efficiency (e.g., checking for duplicates before task creation). These are benign instructional intensifiers and do not attempt to override agent safety guidelines. A surface for indirect prompt injection exists because the agent reads and processes task data from an external backlog, but this is an inherent and necessary component of the task management functionality.\n
- Ingestion points: task_search, task_list, and task_view tools ingest task descriptions and notes from the backlog.\n
- Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore embedded commands when processing backlog data.\n
- Capability inventory: task_create, task_edit, task_complete, and git commit commands are used to act on data.\n
- Sanitization: Absent; the skill does not provide instructions for validating or escaping external content before processing.
Audit Metadata