Tools · Planning and checking

synthesize_evidence

Read-only

Turn raw observations into facts, inferences, gaps, and recommendations.

When to use: Use this after search, connector, or workspace tools when several observations must be distilled.

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 call
{
  "tool": "synthesize_evidence",
  "arguments": {
    "question": "Why do exports time out?",
    "observations": [
      "Timeouts only happen for accounts with more than 200k rows.",
      "The export query scans the whole events table.",
      "Adding an index on events(account_id, created_at) halved the time in staging."
    ]
  }
}

Real output

Question: Why do exports time out?
Facts:
- Timeouts only happen for accounts with more than 200k rows.
- The export query scans the whole events table.
- Adding an index on events(account_id, created_at) halved the time in staging.
Inferences:
- Need more evidence before drawing stronger conclusions.
Gaps:
- Confirm the missing source of truth and unresolved assumptions.
Recommendations:
- Gather the next highest-signal piece of evidence.
- Verify the most important claim before acting on it.

Parameters

NameTypeDescription
observations*arrayFacts, excerpts, or observations to synthesize
questionstringOptional framing question

* required

Plan the work, weigh options, and check results before answering.