feature-planner
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the Bash tool to inspect the local
package.jsonfile for validation scripts such as lint, test, or typecheck and offers to execute them. This execution is scoped to the existing project scripts and is explicitly gated by a request for user confirmation. - [PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection as it captures free-text user responses via the
AskUserQuestiontool and writes them into documentation files. - Ingestion points: User input captured during the interview steps in
SKILL.mdis written to/docs/wip/<feature-name>-plan.md. - Boundary markers: None; the user's raw input is placed into markdown sections without delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill possesses the
Bashtool for command execution andWrite/Edittools for file modification. - Sanitization: No sanitization, escaping, or validation of the interviewed content is performed before it is written to the filesystem.
Audit Metadata