The Agent Skills Directory

[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection because it directs the agent to ingest content from untrusted sources without providing mechanisms to distinguish between data and instructions. 1. Ingestion points: Content is retrieved from Slack messages, emails, Google Drive documents, and external press releases (documented in examples/3p-updates.md and examples/company-newsletter.md). 2. Boundary markers: Absent; there are no instructions or delimiters provided to help the agent isolate the ingested content or ignore instructions within it. 3. Capability inventory: The skill uses tools to read from multiple communication and document platforms and then generates summarized reports. 4. Sanitization: Absent; no validation or sanitization of the retrieved content is required before the agent processes it for summarization.

internal-comms