The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to process and synthesize data from potentially untrusted external sources, which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The skill explicitly processes interviews, surveys, support transcripts, public reviews, social media posts, and forum discussions as outlined in SKILL.md and references/research-modes.md.
Boundary markers: There are no instructions provided to use specific delimiters or to disregard embedded instructions within the processed customer data.
Capability inventory: The skill is designed for planning and synthesizing research into deliverables like personas and coded findings. It does not explicitly invoke high-risk tools like shell execution or network requests within the provided instructions, though it mentions recording results to a 'truth-file'.
Sanitization: The instructions do not mention any sanitization, filtering, or validation steps for the external content before it is processed by the agent.

customer-evidence