The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon data from external sources, including vendor documentation, source files, and command output artifacts, which presents a surface for indirect prompt injection.
Ingestion points: The skill explicitly instructs the agent to read Cited findings from vendor documentation and source files (Rule 1 and Rule 2) in SKILL.md.
Boundary markers: There are no specific instructions or delimiters mentioned to isolate or escape potentially malicious content within the ingested external data.
Capability inventory: The skill utilizes the Read, Grep, and Glob tools for data ingestion, and it has the capability to invoke other agents via the Task tool, which could be exploited if the agent is manipulated by injected instructions.
Sanitization: The instructions do not define any sanitization, filtering, or validation logic for the content retrieved from external sources before it is processed by the agent.

contribution-guardrails