believe
Form synthesized beliefs from multiple facts. Beliefs link to their supporting evidence.
{ "tool": "believe", "arguments": { "belief": "This codebase follows repository pattern", "about": ["node_repo_class_1", "node_repo_class_2"] }}Parameters
Section titled “Parameters”| Parameter | Type | Required | Description |
|---|---|---|---|
belief | string | Yes | The belief statement |
about | string[] | Yes | Node IDs this belief is based on |
confidence | float | No | Confidence score 0-1 (default: 0.7) |
tags | string[] | No | Tags for organization |
Examples
Section titled “Examples”From observed patterns:
{ "tool": "believe", "arguments": { "belief": "Production environment needs higher memory limits than staging", "about": ["node_oom_incident_1", "node_oom_incident_2", "node_staging_success"], "confidence": 0.85 }}Architecture conclusion:
{ "tool": "believe", "arguments": { "belief": "The system uses event sourcing for audit trails", "about": ["node_event_store_fact", "node_replay_capability"], "confidence": 0.9, "tags": ["architecture"] }}When to Use
Section titled “When to Use”- Conclusions synthesized from multiple facts
- Patterns observed across sources
- Architectural or design beliefs
- Positions you’re taking based on evidence
When NOT to Use
Section titled “When NOT to Use”- Single observations (use remember)
- Individual facts (use learn)
- Hunches without supporting facts (use remember)
The belief test: “Based on [these facts], I believe [this conclusion].” If you can’t fill in [these facts], it’s not a belief yet.