The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines a workflow for the agent to ingest and analyze data from external, untrusted sources, including official documentation, package registries, and source code repositories. This behavior creates a potential surface for indirect prompt injection, where malicious instructions embedded in external content could attempt to influence the agent's reasoning or subsequent actions.- Ingestion points: The workflow relies on browsing live external candidates and primary technical sources as described in references/evidence-and-currentness.md.- Boundary markers: While the skill uses a structured solution-decision-template.md for output, it does not specify explicit boundary markers or delimiters (such as XML tags or triple quotes) to isolate untrusted external content within the agent's context.- Capability inventory: The skill itself does not request tool permissions or execute code; it provides guidance for agents that possess research and codebase access capabilities.- Sanitization: The instructions emphasize evidence-backed decision-making but do not provide specific mechanisms for the agent to sanitize or escape data retrieved from external URLs before processing it.

evaluate-existing-solutions