Tools · Planning and checking
decompose_problem
Read-onlyBreak a problem into workstreams, assumptions, risks, evidence needs, and next actions.
When to use: Use this when the task needs visible structured reasoning before execution.
Example
Run for real by scripts/tool-examples.py in a small demo project, with Haiku 5.5 where the tool calls a model. The output is exactly what the tool returned.
The call
{
"tool": "decompose_problem",
"arguments": {
"problem": "Nightly CSV exports time out for large accounts",
"objective": "Exports finish under 5 minutes"
}
}Real output
Problem: Nightly CSV exports time out for large accounts Objective: Exports finish under 5 minutes Workstreams: 1. Clarify scope and target outcome. 2. Gather missing evidence and dependencies. 3. Execute the highest-value actions first. 4. Verify the result against constraints and risks. Assumptions: - The available tools and data are sufficient to make progress. - The task can be advanced through observable intermediate outputs. Risks: - Missing credentials, context, or external dependencies can block execution. - Ambiguous requirements can create wasted work. Evidence Needed: - Confirm the source of truth for important decisions. - Check current workspace state before making irreversible changes. Next Actions: - Pick the first evidence-gathering step. - Execute one verifiable task. - Reassess after each meaningful result.
Parameters
| Name | Type | Description |
|---|---|---|
| problem* | string | Problem or task to break down |
| objective | string | Desired outcome |
| constraints | array | Known limits or requirements |
* required
More in Planning and checking
Plan the work, weigh options, and check results before answering.