The Agent Skills Directory

[SAFE]: The skill performs localized text analysis on a specified wiki directory using standard shell utilities. It adheres to the principle of least privilege by not requesting network access or broad file system permissions beyond the designated wiki and output paths.
[PROMPT_INJECTION]: The skill processes user-generated markdown files in the local wiki/ directory, which creates a surface for indirect prompt injection where content in those files could attempt to influence the agent's behavior.
Ingestion points: The skill reads markdown files across wiki/concepts/, wiki/entities/, and wiki/synthesis/ folders, as well as wiki/log.md and wiki/index.md.
Boundary markers: No explicit delimiters or instructions are used when reading files to ensure the agent ignores embedded commands.
Capability inventory: The skill uses Bash (grep, sed, sort, tail) for text filtering and Read, Write, and Edit for local file operations.
Sanitization: Content from the wiki is used for scoring and ideation patterns without prior sanitization or validation.

second-brain-ideate