The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and summarize large amounts of data from potentially untrusted sources (Slack, Email, Google Drive) where attackers or malicious insiders could place instructions to influence the agent's output.
Ingestion points: examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md all explicitly direct the agent to read from Slack channels, Google Drive documents, and company emails.
Boundary markers: There are no instructions or boundary markers provided to help the agent distinguish between data to be summarized and instructions embedded within that data.
Capability inventory: The agent uses search and read capabilities across communication tools to gather data for summarizing.
Sanitization: No sanitization or validation of the ingested content is mentioned before it is interpolated into the generated reports or newsletters.

internal-comms