The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were detected. The skill's scripts perform local file system operations (reading and writing YAML/Markdown) for repository management using standard system tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled markdown files to build a central index. It implements sanitization in scripts/regenerate-context-index.sh using json.dumps to ensure that extracted fields do not break the YAML structure of the generated index. Mandatory evidence: Ingestion occurs in scripts/regenerate-context-index.sh and validation scripts via .context/*.md file reading; frontmatter is delimited by --- markers; capabilities are limited to local file writes via standard library functions; sanitization is performed via emit_scalar using json.dumps to escape scalar values.

context-index