The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user feedback through the get_feedback_insights tool, which could contain malicious instructions designed to alter the agent's behavior or summaries. Ingestion points: The skill calls get_feedback_insights in Phase 2 of SKILL.md to gather recent user sentiment. Boundary markers: There are no defined delimiters or instructions provided to the agent to treat the feedback data as untrusted or to ignore any embedded commands. Capability inventory: The skill possesses the capability to query broad sets of organizational data, summarize metrics, and recommend concrete actions (e.g., shipping experiments or rolling back deployments). Sanitization: The instructions do not specify any filtering or validation for the content returned by the feedback API.

daily-brief