The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is primarily designed to ingest and process untrusted external data, such as articles, posts, and news items.
Ingestion points: External text sources provided by the user or fetched by adapters as described in the source adapter contract section.
Boundary markers: The skill enforces the use of specific metadata tags like source_kind=shabda and the kalpita (unverified claim) epistemic mode to prevent untrusted data from being treated as facts.
Capability inventory: The skill utilizes the nks_update tool and other graph management tools to store and link data, but does not provide direct shell or arbitrary network access within its own logic.
Sanitization: A mandatory verification by observation step is described to validate claims against reality or independent sources before they are promoted within the system knowledge base.

intake