Start an agent session

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only declare readOnlyHint=false and destructiveHint=false, so the description carries the interesting load and does it well: the call is asynchronous and returns once the prompt is queued, with the reply read from GET /sessions/{session_id}/history/messages. It also discloses the auth asymmetry (OAuth vs API key 400), the visibility/session-reuse behavior with incidentId, and the grant requirements for running the agent and viewing the incident.

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?

Front-loaded with the core action, then progressively layers async behavior, auth, and scoping. Nearly every sentence carries non-redundant information, though the closing grant sentence packs two separate permission requirements into one dense clause.

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?

For a 7-parameter mutation tool with sparse schema documentation and an output schema already covering returns, the description supplies the missing operational context: async contract, where replies are read, credential requirements per mode, session-reuse semantics, and required grants. The only real gap is the tuning parameters (effort, modelId, pipelineEditorContext), which is minor against the overall coverage.

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 only 14%, so the description has to compensate, and it does well for incidentId (ownership, workspace visibility, existing-session absorption) and implicitly for agentId/prompt. But effort, modelId, and pipelineEditorContext are undocumented in both schema and description, leaving three parameters semantically opaque.

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+resource pair ("Start a session with an agent and send it a first prompt") and pins the scope with "first prompt," which cleanly separates it from send_agent_session_message for follow-ups. An agent can identify this as the session-creation entry point without opening the schema.

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?

Gives explicit conditions that select between two modes: without incidentId the session belongs to the caller (OAuth token required, API key gets 400), with incidentId it is incident-scoped and workspace-visible. It also warns that an incident's already-open session absorbs the prompt rather than a new one starting. It stops short of naming a sibling as the alternative for follow-up turns, so 4 rather than 5.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.