Skip to content

Add a fixture-backed end-to-end evaluation command #67

Description

@knytcomics-ui

Category

Intermediate

Summary

Add a deterministic command that performs decode, destination extraction, stub lookup, tier mapping, and label comparison.

Current Behavior

The testkit supplies inputs and scores, while decoding and tier mapping live in sibling repositories. No command validates the combined workflow.

Problem

Individual contracts can pass while the resulting warning is wrong or a destination is missed.

Why This Matters

A product-level smoke test gives maintainers and researchers a reproducible evaluation entry point.

Proposed Scope

Consume a pinned fixture release and report decoded destinations, scores, expected labels, derived tiers, and stage-specific failures.

Acceptance Criteria

  • Every current transaction fixture evaluates end to end.
  • Unknown destinations are distinct from low-risk destinations.
  • Label/tier comparisons are explicit.
  • Failures identify fixture and processing stage.
  • CI runs it on fixture changes.

Technical Considerations

Use the extension decoder, adapter StubOracle, and research tier mapping.

Testing Requirements

Cover valid, unknown, and malformed inputs.

Cross-Repository Impact

All four Gryd Lock repositories.

Out of Scope

Do not implement the live Soroban oracle.

Complexity

Cross-repository orchestration with deterministic inputs.

Impact

High — adds the first product-level regression signal.

Suggested Labels

intermediate, area: testing, area: cross-repo

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions