harness-parity-review

Compare Claude Code and Codex harness capabilities to classify parity drift.

2|1|Updated Mar 30, 2026
One-click install
npx skills add https://github.com/Remaker-Digital/groundtruth-kb --skill harness-parity-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-parity-review
Source: https://github.com/Remaker-Digital/groundtruth-kb/tree/main/.claude/skills/harness-parity-review
Command: npx skills add https://github.com/Remaker-Digital/groundtruth-kb --skill harness-parity-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents role failures caused by capability drift between Claude Code and Codex harnesses, where a role expects certain tools, hooks, plugins, or context but the target harness does not provide equivalent functionality.

Core Features & Use Cases

  • Cross-harness capability inventory: Compares harness identities, role assignments, and a declared capability registry to determine what each harness can actually do.
  • Role-required parity classification: Categorizes each required capability as PASS, DEGRADED, MISSING, STALE, EXTRA, UNSUPPORTED, or OWNER_ACTION_REQUIRED to make remediation concrete.
  • Drift correction workflow: Updates registry entries, pointers, adapters, or generated reports to restore semantic capability equivalence, then verifies the result.
  • Verification-ready reporting: Produces a structured parity report that lists missing role-critical capabilities, degraded fallbacks, stale adapters, undeclared extras, and recommended corrections.

Quick Start

Run the parity check across all harnesses by executing: python scripts/check_harness_parity.py --all --markdown.

Frequently Asked Questions about harness-parity-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What is harness parity verification between Claude Code and Codex?

Harness parity verification detects missing, degraded, or unsupported functionality by comparing role-required capabilities against actual harness-provided surfaces in Claude Code and Codex. It classifies drift across harness identities, role assignments, and capability registry entries to prevent role failures.

How do I detect capability drift between Claude Code and Codex harnesses?

Detect capability drift by executing a parity check script that compares role-required capabilities against actual harness-provided surfaces. The script classifies each capability as PASS, DEGRADED, MISSING, STALE, EXTRA, UNSUPPORTED, or OWNER_ACTION_REQUIRED to make remediation concrete.

How do I generate a verification report for harness capability parity?

Generate a structured parity report by running the parity check script with markdown output across all harnesses. The report lists missing role-critical capabilities, degraded fallbacks, stale adapters, undeclared extras, and recommended corrections for review.

Why does a role fail when assigned to a different harness in Claude Code or Codex?

A role fails due to capability drift, where the role expects specific tools, hooks, or plugins but the target harness does not provide equivalent functionality. Parity verification identifies these missing or degraded surfaces and updates registry entries or adapters to restore equivalence.

What is included in a cross-harness capability inventory for Claude Code and Codex?

A cross-harness capability inventory compares harness identities, role assignments, and a declared capability registry to determine what each harness can actually do. It inventories skills, hooks, plugins, MCP tools, startup context, and permissions across both platforms.

Can I correct drift in harness capability registries without breaking existing role assignments?

Yes, you can correct drift safely by updating registry entries, pointers, adapters, or generated reports to restore semantic capability equivalence. The workflow validates results via the parity check script to ensure existing role assignments remain functional.