The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructs the agent to systematically search and aggregate data from highly sensitive corporate sources, including internal Slack channels, private emails, Google Drive documents, and Calendar events. While the primary purpose is legitimate (creating reports), this capability establishes a data access pattern that could be misused to expose private executive communications or confidential project details if the agent is not strictly constrained.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process untrusted content from company-wide Slack channels and public documents. Maliciously crafted messages in these channels could contain hidden instructions that attempt to influence the agent's summarization, tone, or information selection in company-wide newsletters and FAQs.
Ingestion points: Reads content from Slack messages, Email bodies, Google Drive files, and Calendar attachments (identified in examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md).
Boundary markers: The instructions lack explicit boundary markers or 'ignore' directives to prevent the agent from following instructions embedded within the source data it is analyzing.
Capability inventory: The agent has the capability to read extensive historical data and generate reports that are then distributed via Slack and Email.
Sanitization: There are no mentioned mechanisms for sanitizing or filtering instructions out of the ingested text before processing.

internal-comms