Infrastructure · Data Structures
Result
The output of agentic transformation
Definition
Mechanism
The canonical output of an operation (often a Solution). It encapsulates success/failure status, the artifact produced, and associated MetricReading records such as time taken, resources consumed, and confidence score. It serves as the bridge between execution and evaluation.
Invariants
- Provenance: Every Result must trace to the Task and Solver that produced it.
- Immutability: Once yielded, a Result cannot be modified (only superseded).
- Measurability: Resource consumption must be recorded.
Risks
Failure modes
Partial Result
Output is incomplete but presented as complete.
Hallucinated Result
Output claims to solve the task but doesn't.
Unvalidated Result
Result bypasses checking.
Relationships
Builds on
References
Quantitative value bound to a Metric definition
The fractal unit of completed work
Used by
References
Error correction signal
Specification of desired end state
Mind
Run subtasks simultaneously and aggregate
Mind
Apex triage node of a SolverTree, with Pathway Memory as its compounding learning site
Yields
Formal state machine governing task progression
Society
Discover and invoke external tools via structured capability manifests
Reference
Data schema
{
"type": "object",
"required": [
"status",
"stop_reason"
],
"properties": {
"status": {
"type": "string",
"enum": [
"Success",
"Partial",
"Failure"
]
},
"stop_reason": {
"type": "string",
"enum": [
"Completed",
"Budget",
"Quality",
"Error",
"Timeout"
],
"description": "Why execution terminated"
},
"outputs": {
"description": "Produced artifacts — optional; a failed Result may yield none"
},
"task_ref": {
"type": "string",
"description": "The task this result satisfies"
},
"confidence_score": {
"type": "number",
"minimum": 0,
"maximum": 1
},
"provenance_trace": {
"type": "array",
"items": {
"type": "string"
}
},
"metrics": {
"type": "object",
"description": "MetricReading records associated with the operation"
}
}
}