The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes an automated pipeline for processing untrusted external data (EPUB, PDF, and web articles) and generating structured output written to the filesystem. This represents an indirect prompt injection surface where adversarial content in a source document could influence the agent's actions.
Ingestion points: Source documents are ingested in Phase 1 via PDF tools, EPUB extraction, and web_fetch.
Boundary markers: The instructions lack explicit delimiters or 'ignore instructions' warnings to prevent the agent from following commands embedded within the documents.
Capability inventory: The skill has file-writing capabilities (via put_page into projects/ and concepts/) and network retrieval capabilities (via web_fetch).
Sanitization: No sanitization or content filtering is specified for the ingested text before it is synthesized into recommendations.
[COMMAND_EXECUTION]: The document processing workflow describes the use of CLI tools like pdftotext and parsing libraries like BeautifulSoup. These involve subprocess execution or complex parsing of untrusted binary and markup files, which are standard but notable execution patterns.

strategic-reading