The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions to ingest data from untrusted external sources where content can be controlled by various users.
Ingestion points: Files examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md direct the agent to retrieve information from Slack posts, Google Drive documents, and company emails.
Boundary markers: The instructions do not define delimiters or protective prompts to distinguish between the agent's instructions and the external data being processed.
Capability inventory: The skill uses data retrieval and summarization capabilities to generate company-wide reports and newsletters.
Sanitization: No methods for filtering, escaping, or validating the content pulled from Slack or Google Drive are described.
[DATA_EXFILTRATION]: The skill requires access to sensitive internal data sources, including executive communications and team updates. Although this is the stated purpose of the skill, the processing of this information creates a data exposure surface where internal secrets or sensitive plans could be leaked or misused if the agent's instructions are subverted.

internal-comms