The Agent Skills Directory

[SAFE]: The skill's primary function is to create Markdown-based implementation plans. It does not perform network operations, access sensitive credentials, or execute arbitrary shell commands directly.
[COMMAND_EXECUTION]: The skill mentions the use of development tools like pytest and a sub-skill called git-workflow-preferences. These are invoked as part of a structured developer workflow and are limited to testing and Git management within the project scope.
[DATA_EXFILTRATION]: The skill defines specific file paths for persisting plans (e.g., docs/plans/). These paths are restricted to the local project documentation and do not involve exfiltrating data to external domains.
[INDIRECT_PROMPT_INJECTION]: The skill processes specification files (docs/scope/*.md) as input to generate plans. While this represents an ingestion surface for untrusted data, the skill includes a self-review phase to ensure the generated plan matches the requirements, mitigating basic instruction overrides.

writing-plans