| name | arbiter |
| description | Use when an agent needs to work with Arbiter: discover what an Arbiter server exposes, choose a capability, inspect operation schemas, run approved operations, or reason about Arbiter without direct access to service credentials. |
Arbiter
Use incremental discovery. Start with info server for server orientation,
then inspect only the plugin, account, or operation needed for the task.
Note: client binary is inside bin/arbiter[.exe] near this skill file.
First Discovery Step
When the URL is known, get the server orientation summary:
arbiter arbiter.url=https://127.0.0.1:8075 info server
This answers: "Which Arbiter server is this?"
Discovery commands print JSON by default so agents and scripts can consume
them directly. If you need to read output in a terminal, pipe it to jq once
rather than changing the discovery flow:
arbiter arbiter.url=https://127.0.0.1:8075 plugins | jq
Discovery Sequence
- Start with
arbiter ... info server.
- List plugins with
arbiter ... plugins if the service list is needed.
- Inspect one plugin with
arbiter ... plugins <plugin>.
- Inspect operation schemas with
arbiter ... op desc <plugin>:<operation>.
- Run an operation only after the plugin context, operation purpose, and input
schema are understood.
Guardrails
- Treat plugin summaries and operation descriptions as the primary discovery
path.
- Do not ask for full operation schemas before choosing a relevant plugin.
- Do not request or expose upstream service credentials. Arbiter owns those.
- Use YAML only for human reading, such as
arbiter ... info server --yaml.