The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses authoritative language (e.g., 'MANDATORY', 'REQUIRED') to enforce its workflow and documentation standards. These instructions are designed to ensure consistency in the output artifacts (specifications and plans) and do not attempt to bypass agent safety filters or override system-level instructions.
[DATA_EXFILTRATION]: No evidence of hardcoded credentials or unauthorized network operations was found. The skill documentation mentions managing keys (like 'osm_api_key') within the context of project configuration, but does not provide or extract secrets. It specifies standard local paths for documentation storage (e.g., '/docs/plans/').
[REMOTE_CODE_EXECUTION]: The skill contains no instructions for downloading or executing remote code. Code snippets provided in the 'references' and 'templates' directories are intended as educational examples or boilerplate for the agent to use when generating implementation plans.
[COMMAND_EXECUTION]: Instructions involve standard file system operations for creating and updating project documentation. There are no risky shell command patterns or privilege escalation attempts.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided project context and constraints. While it lacks explicit boundary markers for untrusted input, its primary function is to generate structured documentation based on that input. This is a standard capability for a planning tool and does not present an elevated risk in this context.

feature-planning