The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external research briefs which can trigger automated capabilities including file generation and web-based activity.
Ingestion points: Accesses files within the literature-reviews/ directory, specifically evidence-brief.md, sources.md, and literature-review.md.
Boundary markers: The instructions do not specify the use of clear delimiters or instructions to ignore embedded commands within the ingested research files.
Capability inventory: Includes the ability to write to the file system (DOCX, PPTX, PDF) via the ane_package.reporting module and perform network operations using WebSearch and WebFetch.
Sanitization: The skill includes a URL audit step to verify link integrity and resolve redirects, but it lacks general content sanitization to prevent the interpretation of malicious instructions within the source text.
[DATA_EXFILTRATION]: The skill performs network operations to external, non-whitelisted domains to verify cited URLs and also accesses configuration files in the user's home directory.
The Session-time URL audit step utilizes WebFetch and WebSearch to reach out to URLs found in research briefs for verification.
The skill reads ~/.claude/CLAUDE.md, a configuration file located in the user's home directory.
[COMMAND_EXECUTION]: The skill utilizes the ane_package.reporting suite, a vendor-owned resource, to generate document exports.
Specifically invokes pptx_export.write_pptx_deck, word_export.write_word_report, and pdf_export.write_pdf_report.
[EXTERNAL_DOWNLOADS]: The citation verification process involves fetching content from various external URLs discovered in the research briefs using WebFetch.

learning-product