The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a workflow for ingesting untrusted data from external sources and processing it using an LLM, creating a vulnerability surface for indirect prompt injection.
Ingestion points: The ingestDocument and checkSourceFreshness functions (found in references/curation.md) ingest content from raw documents and remote URLs.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the processing logic.
Capability inventory: The skill utilizes network operations (fetch) to retrieve content and passes data to an LLM via the autoTag function.
Sanitization: While basic cleaning functions like stripHtml and normalizeMarkdown are implemented, they are designed for formatting rather than security-focused sanitization against adversarial inputs.
[EXTERNAL_DOWNLOADS]: The skill contains reference code for fetching content from external URLs to perform freshness checks and content ingestion.
Evidence: The checkSourceFreshness function in references/curation.md uses the fetch API to perform HEAD and GET requests on external URLs provided in SourceRecord objects.

knowledge-base-manager