The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the use of the bt CLI for running local evaluations (bt eval), executing SQL queries (bt sql), and configuring environment setup (bt setup). These commands are core to the Braintrust platform's functionality.
[EXTERNAL_DOWNLOADS]: The skill includes documentation for the bt docs fetch command, which retrieves technical references and workflow guides from the official braintrust.dev domain to provide up-to-date context for the agent.
[DATA_EXFILTRATION]: The skill enables the transmission of project logs, traces, and metadata to the Braintrust service at braintrust.dev. This process utilizes standard secure credential management, including system keychains and restricted file permissions for local secrets.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection as it ingests data from external logs and traces, which could theoretically contain instructions intended to influence the agent.
Ingestion points: Data enters the agent context through bt view logs and bt view trace commands, as well as fetched documentation.
Boundary markers: The instructions do not define specific delimiters or boundary markers to isolate the ingested external data.
Capability inventory: The agent has the ability to execute shell commands and run local code evaluations via the bt CLI.
Sanitization: The skill does not explicitly describe sanitization or filtering of the project data retrieved from the external service.

braintrust