build
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and acts upon data from potentially untrusted external sources like plan.md and GitHub issues.
- Ingestion points: plan.md, masterplan.md, and GitHub issue descriptions.
- Boundary markers: No explicit delimiters are used to isolate content.
- Capability inventory: Execution of local scripts (plan-refresh.sh) and invocation of task-building tools (section-builder).
- Sanitization: No sanitization of ingested task text is performed.
- [COMMAND_EXECUTION]: The skill executes local utility scripts, such as .agents/tools/plan-refresh.sh, to manage task synchronization within its environment.
- [EXTERNAL_DOWNLOADS]: Connects to GitHub to fetch project issue data. GitHub is a well-known service and this interaction is standard for the skill's purpose.
Audit Metadata