The Agent Skills Directory

[SAFE]: The skill implements a mandatory Sanitization Policy and Pre-Capture Check that strictly prohibits the storage of environment variables, API tokens, passwords, and private identifiers in project wiki artifacts.- [COMMAND_EXECUTION]: The SKILL.md provides instructions for running local validation and linting scripts located in the platform's system directories ($CODEX_HOME/skills/.system/), which is standard for development workflows.- [PROMPT_INJECTION]: The skill ingests data from local documents and user-provided artifacts to create persistent memory, representing an indirect prompt injection surface. Ingestion points: Reads from existing documents via tools like obsidian-wiki-ingest. Boundary markers: None explicitly defined for the generated wiki artifacts. Capability inventory: Writing durability facts and context packs to the local file system. Sanitization: Mitigated by mandatory secret-redaction and fact-labeling protocols (User-provided, Source-backed, etc.).

project-wiki-manager