feature-planner
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a Bash command block designed to log metadata to a local file (
~/.mystack/timeline.jsonl). This command uses shell interpolation$(...)to fetch the current Git branch and repository name. While this allows the agent to automate project tracking, it relies on the local Git environment being in a safe state. - [PROMPT_INJECTION]: The skill ingests arbitrary user input describing feature ideas and interpolates it into markdown documents and shell commands. This creates a surface for indirect prompt injection where malicious text provided as a 'feature description' could attempt to influence the agent's behavior during the documentation or logging phases. However, the skill's rigid phase-based structure and 'one question at a time' rule significantly mitigate the risk of the agent deviating from its primary task.
Audit Metadata