The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze "historical patterns," "current signals," and "weak signals of change," which are typically sourced from external, untrusted content (e.g., web searches or documents). This creates a vulnerability surface where malicious instructions could be embedded in the data being processed.
Ingestion points: The skill instructions in SKILL.md specify reviewing signals and patterns from the context manager.
Boundary markers: There are no explicit boundary markers or instructions to treat external signal data as untrusted content.
Capability inventory: The skill does not define custom scripts, but it leverages the agent's default capabilities to analyze and summarize external information.
Sanitization: No mechanisms for sanitizing or filtering instructions out of the trend data are provided.

trend-analyst