harness-eval

A testing tool for checking whether an agent harness—the files and procedures that guide coding agents—is complete and produces good work. It uses structure checks, content checks, and isolated tasks judged against alternatives.

In plain words
What is it for?
Use it to audit required files, validate instruction and configuration content, run fixture-based agent evaluations, and compare two harness variants.
Why use it?
It shows whether the harness is installed correctly and whether its instructions influence agent behavior. This makes weaknesses measurable before the harness is changed.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/doidor/agentrig/harness-eval
Any agent
npx skills add doidor/agentrig --skill harness-eval
Clone the repo
git clone --depth 1 https://github.com/doidor/agentrig

Made for: Claude Code, Codex.

Per session 48 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,196 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00048 $0.01196
Opus 5 $0.00024 $0.00598
Sonnet 5 $0.00010 $0.00239
Haiku 4.5 $0.00005 $0.00120

Measured 2d ago against content hash 7c0df358b52f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

harness-eval scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/harness-eval/SKILL.md · 93 lines

How it starts

The opening of the file, as written. The whole thing — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.

harness-eval (principle 6 — evaluate the harness itself)

A harness you cannot measure is a harness you cannot improve. This skill scores the harness on three complementary layers and writes results to .agentrig/eval/results/ (validated on write and on read; never hand-edit JSON).

Layer A1 — install completeness (deterministic, no model)

Every canonical artifact present at the path the manifest declares.

node .agentrig/eval/static-audit.mjs --json   # Install Completeness %

Layer A2 — quality probes (deterministic, no model)

Cheap content sanity: YAML parseable, no unfilled {{PLACEHOLDER}} in AGENTS.md, every skill has the required frontmatter, axes.json has an issue code per axis, developer/reviewer model families differ (not just the model id strings).

A1 + A2 are what CI gates on. Both surface in the same --static report under "Layer A1" and "Layer A2" sections.

Layer B — dynamic behavioral eval (agentic, independent judge, fixture-based)

For each scenario in .agentrig/eval/scenarios/*/:

  1. Seed a throwaway worktree from scenarios/<id>/fixture/ (or baseline/+change/ for review scenarios).
  2. Producer model runs in that worktree against scenarios/<id>/prompt.md. For --variant harness, the AgentRig harness is staged into the worktree first; for --variant baseline, the agent runs bare.
  3. Oracle (scenarios/<id>/oracle.yml) deterministically scores the hard axes (correctness, tests, scope, regression_risk, …) by running commands / inspecting the diff. No LLM.
  4. Judge model — explicitly a different family from the producer — runs in a separate provider.startConversation() call in its own cwd containing only prompt.md, diff.patch, transcript.md, oracle.json, and judge_brief.md. It does NOT see the producer worktree or reasoning trace. It writes <artifactsDir>/<scenario>.trial<N>.judge.json; the orchestrator reads, validates, and persists via score.mjs save.

Read the full file on GitHub · 93 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 93 lines · 48 tokens per session scan A 7c0df358b52f

Subscribe to this mod's changes

harness-eval is a skill published in the GitHub repository doidor/agentrig (5 stars, last pushed 1mo ago), licensed MIT. It adds 48 tokens to every session and 1,196 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

brainstorming

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

obra/superpowers · 37 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

agent-host-chat-contributions

Build and review cross-cutting agent-host chat behavior through lifecycle contributions. Use when adding turn lifecycle side effects, prompt or context injection, restored-history transformation, protocol-action observation, or when reviewing changes that add code to AgentSideEffects or AgentService.

microsoft/vscode · 56 tokens

auto-perf-optimize

Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.

microsoft/vscode · 62 tokens