The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill explicitly instructs the agent to ingest and summarize data from untrusted multi-user sources, which is a primary vector for indirect prompt injection.
Ingestion points: examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md tell the agent to search Slack channels (prioritizing high-engagement posts), emails, and Google Drive documents.
Boundary markers: Absent; there are no instructions to treat retrieved content strictly as data or to ignore embedded commands.
Capability inventory: The skill contains no executable code but leverages the agent's built-in tools for reading internal data.
Sanitization: Absent; no guidance is provided to the agent for validating or filtering content retrieved from external sources before inclusion in communication drafts.

internal-comms