The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @onequery/cli package from npm, which is the expected tool for interacting with the OneQuery service.
[COMMAND_EXECUTION]: The agent uses the onequery CLI to perform authentication, organization lookup, and read-only SQL queries.
[SAFE]: The skill incorporates safeguards against Indirect Prompt Injection (Category 8) by instructing the agent to treat CLI output as data rather than instructions.
Ingestion points: CLI output from onequery query exec and metadata commands.
Boundary markers: Specific guardrail in SKILL.md to ignore instructions in output.
Capability inventory: Shell access to the onequery binary.
Sanitization: Agent-led separation of data and control logic per instructions.
[SAFE]: Query safety is maintained by requiring validation and applying strict resource limits like --max-rows and --max-bytes.

onequery-cli