The Agent Skills Directory

[SAFE]: The skill serves as a template and instructional guide for creating VISION.md files. It focuses on clarifying project goals, target audiences, and non-goals through a structured drafting process.
[PROMPT_INJECTION]: No malicious injection patterns or attempts to override AI safety constraints were identified. The skill includes defensive instructions that explicitly separate project intent from operational agent rules to prevent instruction confusion.
[DATA_EXFILTRATION]: The skill does not perform network requests or access sensitive files. It operates entirely within the scope of project documentation and metadata.
[COMMAND_EXECUTION]: There are no shell commands, script executions, or subprocess invocations. The instructions are purely for natural language processing and document formatting.
[REMOTE_CODE_EXECUTION]: The skill does not download external code or define remote dependencies. It relies on standard markdown templates.
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private tokens are present in the skill's content or metadata.

vision-forge