The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process data from external, potentially untrusted sources like Slack, Email, and Google Drive.
Ingestion points: Instructions in examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md direct the agent to read content from Slack, Google Drive, Email, and Calendars.
Boundary markers: Absent. The skill does not provide instructions to separate retrieved data from processing instructions or to ignore commands within the data.
Capability inventory: Uses tools to read internal communications for the purpose of text generation and summarization.
Sanitization: Absent. There is no mention of filtering or escaping content fetched from tools.

internal-comms