agent-rack-tool-selection
Delegating any task to a sub-agent through agent-rack
How to use it
agent-rack-tool-selection is exposed by the Agent Rack MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the agent-rack-tool-selection tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Agent Rack
npx agent-rack setupOther tools in Agent Rack (39)
boolean
string
No parameters. Lists every configured agent and whether its binary is on $PATH.
Same execution parameters as agentrun (agent, prompt required; workspace, mode,
An agentreview call returns
Map of agent id → { name, command, args, transport, env, description, model, inheritEnv }
Absolute directory paths agents may be launched in. Every tool call resolves symlinks and validates against this list before any subprocess spawns. See the caveat under Security model — this constrains the working directory, not everything the process can reach.
boolean
string
number
When transport is stdio (the default), also open a loopback SSE listener on the same session state, so agent-rack session status/tail and the dashboard can observe sessions created over the stdio connection with no separate server. Default true; same bearer-token protection as security.requireSseAut
string
Agents
number
/lineend may be 0 for whole-file, deleted-file, or architectural findings.
opencode
string
string
string
agent-rack install --target opencode
HTTP port when transport is sse, or for the sidecar below (default 8987)
string
claude, codex
working-tree \
read-only \
Cap on simultaneously running background sessions (default 5)
Require a bearer token on the SSE transport (default true). See SSE authentication.
Strip env vars matching credential patterns before spawning agents (default true). Prefer per-agent inheritEnv.
string
number
number
stdio (default, for per-client local IDE integration) or sse (localhost HTTP-SSE, for a shared server and dashboard)
string