The Agent Skills Directory

[SAFE]: No malicious instructions, obfuscation, or unauthorized data exfiltration patterns were found within the skill's own instructions. The tool uses standard utilities like pandoc and TreeSitter for document processing.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because it is designed to ingest and summarize untrusted user documents. * Ingestion points: Source documents are built into a manifest and processed in Step 2 and Step 5. * Boundary markers: The instructions do not define explicit boundary markers or 'ignore' delimiters for the ingested content. * Capability inventory: The skill possesses file-writing capabilities (generating Markdown and data files) and performs multi-step recursive reasoning using AI models (Steps 7, 13, and 15). * Sanitization: No explicit sanitization or filtering of the source document content is performed before it is passed to AI models for summarization.

handbook-writer