grok_agent_startStart delegated coding tasks asynchronously by sending a versioned, typed task packet to the Grok CLI over stdio, WebSocket, or legacy transport. Use auto for stdio only.
Asynchronously start a versioned typed task packet on an explicit legacy|stdio|websocket transport. auto means stdio only.
| Name | Required | Description | Default |
|---|---|---|---|
| lane | No | ||
| task | Yes | ||
| transport | No | stdio |
Changes observed during successful MCP inspections.
v0.33.0Input schema / properties / task / properties / timeout_seconds / maximumPrevious value: -3600New value: +86400v0.26.1Input schema / properties / task / properties / mount_pathsAdded value: +{
+ "items": {
+ "maxLength": 1024,
+ "type": "string"
+ },
+ "maxItems": 16,
+ "type": "array"
+}Input schema / properties / task / properties / review_laneAdded value: +{
+ "maxLength": 96,
+ "minLength": 1,
+ "type": "string"
+}v0.4.1Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With annotations already declaring non-read-only, open-world, non-idempotent behavior, the description adds the genuinely useful async execution nature and transport semantics. It does not disclose required permissions, whether a job handle/ID is returned, what happens on task-schema violation, or any timeout/cancellation implications.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, purpose front-loaded, no filler. The second sentence is a useful caveat about the transport default rather than padding, though the dense jargon slightly works against fast comprehension.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
A mutation tool with a deeply nested, fully undocumented task payload, no output schema, and 0% parameter coverage needs far more from the description. It omits the shape of the task packet, the meaning of lane/role, the return value, and the follow-up lifecycle, leaving an agent unable to construct a valid call from the description alone.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0% across three parameters, including a large nested 'task' object with 15+ undocumented fields (role, lane, permission_profile, acceptance_criteria, etc.). The description only clarifies one enum value by noting 'auto means stdio only'; 'lane' and the entire task packet contract are left unexplained.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb and resource ('start a versioned typed task packet'), which distinguishes it from the status/cancel/review siblings. However, 'task packet' is opaque jargon an agent may not decode, and it never distinguishes itself from close siblings like grok_agent_execute or grok_delegate_start.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No when-to-use or when-not-to-use guidance. There is no mention of how this differs from grok_agent_execute, grok_agent_fix, or grok_delegate_start, nor any prerequisite or lifecycle context (e.g., that a started job is later polled via grok_agent_status). The only usage hint is a transport default note.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.