The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection. \n
Ingestion points: Data is retrieved from Slack messages, Gong transcripts, Jira tickets, and HubSpot notes via tools like gong-calls and hubspot-records in SKILL.md.\n
Boundary markers: The instructions do not define clear delimiters or explicit warnings for the agent to ignore instructions within the retrieved data during synthesis.\n
Capability inventory: The agent can perform save-analysis to persist new instructions, create-extension to generate code/UI, and provider-api-request for arbitrary network calls as described in SKILL.md.\n
Sanitization: There is no defined process to sanitize or escape the content retrieved from external sources before it is interpolated into the instructions for saved analyses.\n- [COMMAND_EXECUTION]: The skill uses the create-extension action to dynamically generate and deploy new code or user interface components. It also utilizes provider-api-request to perform arbitrary API operations against configured third-party services.\n- [EXTERNAL_DOWNLOADS]: The skill is designed to interact with and pull data from a wide variety of external services, including HubSpot, Gong, Slack, BigQuery, GitHub, Jira, and Stripe. While central to its purpose, this extensive integration increases the platform's exposure to potentially malicious content hosted on these services.

adhoc-analysis