The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted user-generated content (session logs, issue titles, and workflow descriptions) that could be used for indirect prompt injection attacks.
Ingestion points: As noted in the 'Trust boundary' section of SKILL.md, data returned by the bd CLI or the bitdrift API may contain arbitrary text.
Boundary markers: SKILL.md contains a specific 'Trust boundary' section that instructs the agent to treat retrieved content as data rather than instructions and to ignore any attempts to override agent behavior.
Capability inventory: The skill enables the agent to execute various bd CLI commands for platform management and use curl for direct API calls.
Sanitization: The instructions emphasize using -o json and --jq for structured parsing, which helps prevent the accidental execution of instructions embedded in text logs.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the bd CLI tool via Homebrew from the vendor's repository (bitdriftlabs/bd).

bd-cli