The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes untrusted external content from financial news, corporate filings, and announcements (e.g., via longbridge news and longbridge filings subcommands described in references/corporate-events.md and references/smallcap-growth.md). This ingestion of external data constitutes a surface for indirect prompt injection, as malicious instructions could theoretically be embedded in these documents to influence the agent's behavior.
Ingestion points: External data enters the context via longbridge news and longbridge filings subcommands.
Boundary markers: The instructions do not define explicit boundary markers or delimiters to isolate untrusted external content.
Capability inventory: The skill enables the agent to execute a wide variety of longbridge CLI subcommands for financial data retrieval and analysis across all reference files. No arbitrary shell execution or file-write capabilities were detected in the skill instructions.
Sanitization: There are no instructions for sanitizing, escaping, or validating the content retrieved from external news or regulatory filing sources.

stock-fundamentals