The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data (Substack posts, Twitter threads, and research reports) into a local "knowledge directory" which is subsequently scanned and loaded during the analysis phase.
Ingestion points: Files are parsed and saved to subdirectories within the user-defined knowledge path (documented in references/commands/import.md). These files are then loaded by the analysis command (documented in references/commands/analysis.md).
Boundary markers: The instructions for loading this content into the agent's context do not include the use of specific delimiters or instructions for the agent to ignore potentially malicious embedded instructions within the ingested data.
Capability inventory: The skill has access to financial APIs (Unusual Whales, Funda, TradingView) and the ability to read and write local files.
Sanitization: There is no explicit sanitization or filtering mechanism described for the ingested external content before it is processed by the agent.

trade