The Agent Skills Directory

[SAFE]: The skill provides a structured methodology for technical planning and contains no malicious instructions or patterns. It operates locally within the project directory and does not request network access or sensitive system credentials.
[COMMAND_EXECUTION]: The skill uses local utility scripts (scripts/check_paths.sh and scripts/check_progress.sh) for plan validation. These scripts are implemented securely, utilizing strict regular expression filtering to ensure that only valid, safe file paths are processed, effectively preventing any command injection from plan content.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests project data and user requirements to generate plans, it incorporates architectural quality gates and fact-checking requirements that reduce the risk of following malicious instructions embedded in codebase documents.

dev-plan