harness-eval

This skill should be used when the user asks to "test the harness", "run integration tests", "validate features with real API", "test with real model calls", "run agent loop tests", "verify end-to-end", or needs to verify OpenHarness features on a real codebase with actual LLM calls.

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

85%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-structured eval skill with concrete code, explicit validation loops, and clean reference splitting. The main weakness is organizational redundancy that inflates length without adding new information.

Suggestions

Consolidate the repeated max_turns guidance (currently in Core Principles, Section 5, Section 7, and Common Pitfalls) into a single authoritative spot and cross-reference it.

Merge the 'do not abandon long tasks' advice, which appears both in Section 5 ('keep polling... do not abandon a run after the first long pause') and Common Pitfalls, into one location.

Dimension	Reasoning	Score
Conciseness	It avoids explaining concepts Claude already knows, but repeats guidance across sections (max_turns lowering warnings in Core Principles, Section 5, Section 7, and Common Pitfalls; 'do not abandon long tasks' stated twice), so it could be tightened rather than earning a lean 3.	2 / 3
Actionability	Provides executable bash (git clone, exports, apt-get, srt --version), real Python snippets (sandbox settings, engine/collect pattern), concrete test-file paths, and per-scenario success criteria that are largely copy-paste ready.	3 / 3
Workflow Clarity	A clear 7-step sequence with explicit validation checkpoints (sandbox smoke check, assertion examples), a failure-classification feedback loop ('classify before changing code'), and a feature coverage checklist.	3 / 3
Progressive Disclosure	Heavy code templates and the feature matrix are split into real, one-level-deep reference files (test-patterns.md, feature-matrix.md) that are clearly signaled both inline and in a dedicated Additional Resources section.	3 / 3
	Total	11 / 12 Passed

Description

90%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, third-person description with explicit, natural trigger terms and a clear niche. Its only weakness is that the 'what' is a single capability statement rather than a list of concrete actions.

Dimension	Reasoning	Score
Specificity	It names the domain ('verify OpenHarness features on a real codebase with actual LLM calls') but states only one capability rather than listing multiple concrete actions like the PDF example; not vague enough for 1, not a multi-action list for 3.	2 / 3
Completeness	Explicitly answers both 'what' (verify OpenHarness features on a real codebase with actual LLM calls) and 'when' (the quoted trigger phrases plus 'when the user asks to...'), with explicit trigger guidance that lifts it above 2.	3 / 3
Trigger Term Quality	Quotes several natural phrases a user would actually say ('test the harness', 'run integration tests', 'verify end-to-end', 'test with real model calls'), giving good coverage of common variations.	3 / 3
Distinctiveness Conflict Risk	Tied to a specific product (OpenHarness) with distinctive triggers like 'test the harness', giving it a clear niche unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: HKUDS/OpenHarness
Path: .claude/skills/harness-eval/SKILL.md
Commit: 9b2efd7

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.