⚠️ Draft — Non-authoritative content (noindex)
INFORMATIVEDRAFT
Truth Source: Repository schemas and tests are authoritative.
Golden Test Suite Overview
1. Philosophy
The Golden Test Suite serves as a reference evidence artifact for the MPLP v1.0 Protocol. It provides a standardized set of input/output pairs that demonstrate the expected behavior of a conformant runtime.
It is not a certification tool or a compliance enforcement mechanism. It is a collection of data fixtures that implementers can use to verify their own alignment with the protocol.
2. Evidence Scope
The suite provides evidence for the following protocol areas:
| Flow ID | Name | Evidence Focus |
|---|---|---|
| FLOW-01 | Single Agent – Happy Path | L1 Core (Context/Plan) Schema Alignment |
| FLOW-02 | Single Agent – Large Plan | L1 Core (Trace/Constraints) Invariant Adherence |
| FLOW-03 | Single Agent – With Tools | L2 Coordination (agent_role/Extension) |
| FLOW-04 | Single Agent with LLM Enrichment | L3 Runtime (AEL) Plan Generation |
| FLOW-05 | Single Agent with Confirm Required | L2 Coordination (Confirm/Trace) Approval Flow |
3. Artifact Structure
The suite consists of:
- Fixtures: JSON files defining inputs (Context, Plan) and expected outputs (Trace, Events).
- Registry: A catalog of defined test scenarios (see Golden Flow Registry).
- Format Specification: Definitions of the fixture directory structure (see Golden Fixture Format).
4. Intended Audience
This documentation is intended for:
- Protocol Implementers: To understand the expected data shapes for specific scenarios.
- SDK Maintainers: To verify SDK output against standardized fixtures.
- CI Integrators: To understand the structure of the test artifacts.