Skip to content

[Extension]: Evaluator Contract #4403

Description

@tbitcs

Extension Submission

Extension Metadata

  • ID: evaluator
  • Name: Evaluator Contract
  • Version: 1.0.0
  • Description: Standard evaluator result contract for evidence, provenance, uncertainty, and recovery — a provider-neutral protocol for extensions that evaluate artifact quality between Spec-Driven Development phases
  • Author: ElectroHire
  • License: MIT

Links

Requirements

  • Spec Kit Version: >=1.0.0
  • Tools: None required

Provides

  • Commands: 4 (speckit.evaluator.run, .compose, .report, .route)
  • Hooks: 4 (after_specify, after_plan, after_tasks, after_implement)

Tags

evaluator, evidence, provenance, quality, governance, compliance, workflow, model-routing, portfolio

Key Features

  • JSON Schema for evaluator results (6 outcomes, 14 finding kinds, 5 evidence kinds)
  • Deterministic composition with contradiction detection (strict/majority/optimistic)
  • Model routing: recommends budget/standard/premium tier per phase
  • All 5 report formats: terminal, markdown, JSON, CI annotations, gate
  • 3 parity scripts: Python, Bash, PowerShell
  • Quick-start demo: python examples/demo.py
  • 70 tests with 0 regressions against spec-kit test suite
  • Token-economic benchmark suite with Monte Carlo simulation

Testing Confirmation

  • Tested locally with uv run specify --help
  • Ran existing tests with uv sync && uv run pytest
  • Tested with a sample project (install → verify → uninstall)

Refs: #4290, #4390

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions