plan-feature
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows secure design principles by explicitly instructing the agent not to include deployment or shared environment verification steps in the generated plan unless the user gives an explicit 'yes' via the questioning tool.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied feature descriptions or existing plan files to generate work phases, which creates a potential surface for indirect prompt injection. However, the risk is considered low given the intended use case of professional feature planning and the requirement for user review. 1. Ingestion points: Feature descriptions or plan file paths provided by the user (SKILL.md). 2. Boundary markers: No explicit delimiters or boundary instructions are mandated for the user-provided content. 3. Capability inventory: Performs local file reads to research code patterns, writes markdown plan files to the 'plans/' directory, and interacts with the user via a questioning tool. 4. Sanitization: No specific sanitization or filtering of user-provided descriptions is defined in the instructions.
Audit Metadata