Skip to content

Opt Quality

opt_quality — mvp.opt_quality

Cluster: Uncategorised | Type: component | MCP Tools: 5

Overview

Quality gap detection and quality improvement intervention recommendations. Analyzes workflow traces to identify low-quality steps and recommends verification, grounding, model upgrades, retries, ensembles, and context enrichment interventions.

Example:

from mvp.opt_quality import OptQualityBlock, OptQualityInput

block = OptQualityBlock()
result = block.infer(OptQualityInput(
    op="analyze",
    trace={
        "trace_id": "trace-1",
        "workflow_name": "demo_pipeline",
        "timestamp": "2026-05-21T00:00:00Z",
        "steps": [
            {"step_index": 0, "component_name": "llm_router",
             "operation": "infer", "confidence": 0.55, "latency_ms": 1200.0},
        ],
    },
))
# result.value -> OptQualityOutput with detected quality bottlenecks

Production Caveat

Heuristic reliability signal, not a correctness guarantee

opt_quality identifies quality risks from workflow traces and recommends likely reliability improvements. Its recommendations are heuristic and should be treated as triage guidance, not proof that a workflow is correct or that an intervention will improve production behavior. For high-stakes, regulated, or unattended workflows, validate changes with held-out tests, passing verification steps, and human review before deployment.

Public API

OptQualityBlock(AIBlock[OptQualityInput, OptQualityOutput, dict])

Quality-focused optimisation block.

Field Type Default
name str 'opt_quality'
resource_bounds ResourceBounds \| None None
usage ResourceUsage field(default_factory=ResourceUsage)
state dict \| None field(default_factory=dict)

Methods:

infer(data: OptQualityInput) -> Result[OptQualityOutput]

OptQualityInput(BaseModel)

Field Type Default
op Literal['analyze', 'recommend', 'analyze_and_recommend', 'estimate_improvement', 'list_patterns', 'capabilities'] 'analyze_and_recommend'
trace dict \| None None
bottlenecks list[dict] \| None None
interventions list[dict] \| None None
quality_threshold float 0.8

OptQualityOutput(BaseModel)

Field Type Default
op str required
report dict \| None None
estimated_improvements list[dict] \| None None
patterns_payload dict \| None None
success bool True
error str ''
degraded bool False
degradation_reason str \| None None
completion_state Literal['verified', 'qualified-draft', 'blocked-escalated'] 'qualified-draft'
warning_card dict[str, Any] \| None None
evidence dict[str, Any] Field(default_factory=dict)
request_id str \| None None
task_id str \| None None
run_id str \| None None

MCP Tools

Operation Source
analyze_quality opt_quality_mcp
recommend_quality opt_quality_mcp
analyze_and_recommend_quality opt_quality_mcp
list_quality_patterns opt_quality_mcp
quality_capabilities opt_quality_mcp

Operations

Op Description
analyze Detect quality gaps in a workflow trace
recommend Generate quality improvement interventions for given bottlenecks
analyze_and_recommend Combined analysis and recommendation in one call
estimate_improvement Estimate quality improvement of proposed interventions
list_patterns Return deterministic pattern and skill catalog introspection
capabilities Return analyzers, component sets, dispatch table, thresholds, maturity, and limitations

MCP Surface

The additive opt_quality_mcp facade exposes analyze_quality, recommend_quality, analyze_and_recommend_quality, list_quality_patterns, and quality_capabilities. MCP responses carry the same completion_state, warning_card, evidence, request_id, task_id, and run_id envelope fields as the local block.

Gateway Tool

This component powers the quality axis of the optimization_report gateway tool (Premium tier).

See also