xp-harness: Skill for Claude Code

.claude/skills/harness-verification/SKILL.md

harness-verification is a skill for Claude Code from sei-newbear/xp-harness. It costs 185 tokens per session (11,342 once invoked), scanned A, original, MIT.

A verification method for coding-agent skills, subagents, and instructions. It checks whether they start when expected and behave as intended by examining recorded sessions.

In plain words
What is it for?
Use it to set up a test state, run the agent, inspect its transcript, and clean up afterward. It also helps assess real sessions against explicitly defined expectations using evidence rather than scores.
Why use it?
Configuration changes can look correct without actually running correctly. This provides a repeatable way to observe the agent, either in a test environment or in a real session.

Skill for Claude Code

Written for Claude Code: installed under .claude/. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions subagents.

This is sei-newbear/xp-harness's own configuration. It tells Claude Code how to work on xp-harness itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything xp-harness configures →

Reuse

Borrowing it

Nothing to install: this file belongs to sei-newbear/xp-harness. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/sei-newbear/xp-harness/main/.claude/skills/harness-verification/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/sei-newbear/xp-harness

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for harness-verification

README.md
[![agentmods](https://agentmods.dev/badge/skills/sei-newbear/xp-harness/harness-verification.svg)](https://agentmods.dev/skills/sei-newbear/xp-harness/harness-verification)
Your own site
<a href="https://agentmods.dev/skills/sei-newbear/xp-harness/harness-verification"><img src="https://agentmods.dev/badge/skills/sei-newbear/xp-harness/harness-verification.svg" alt="Measured on agentmods" height="20"></a>
Per session 185 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 11,342 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00185 $0.11342
Opus 5 $0.00093 $0.05671
Sonnet 5 $0.00037 $0.02268
Haiku 4.5 $0.00018 $0.01134

Measured 7d ago against content hash 5e6889417856, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

harness-verification 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 7d ago.

The scan reads SKILL.md. This mod also ships 19 executable files (scripts/analyze-session.py, scripts/claude-launcher.sh, scripts/find-verifications.sh, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

.claude/skills/harness-verification/SKILL.md · 246 lines

How it starts

The opening of the file, as written. The whole thing — 246 lines — stays where its author put it; the contents beside it link to each section on GitHub.

harness 検証 — skill / subagent / instruction の発火・振る舞いの事実確認

なぜこの skill があるか

skill / subagent / instruction を改修しても、それが実際に発火するか・意図どおり振る舞うかは transcript を観測しないと分からない。検証の手段は、対象が配布物か改修者向けかで決まる — 配布物 (.apm/ 配下) は sandbox に deploy されるので sandbox で走らせて見る。改修者向け (project root の CLAUDE.md や .claude/ 直置きの skill) は sandbox に deploy されないので、本番の実運用セッションの解析でしか観測できない。この検証のやり方 (sandbox の作り方、起動の仕方、transcript の場所と解析方法) はノウハウの塊で、skill 化しないと検証のたびにゼロから再発明することになる。

ここで言うのは「検証」= 仕組みが意図どおり動いたか(sandbox: 的は改修が定める=既知)・あるべき振る舞いができているか(本番: 的=理想を自分で立てて照らす)の確認。核は「できた / できていない」の事実確認。加えて本番セッションでは自走の 良し悪し(止まり方が妥当だったか・判断の質など)まで踏み込んでよい ── ただし二つを守る: 評価は必ず 証拠に接地する(印象・自己分析でなく生 transcript で裏を取る)、点数化・ランク・ベンチには逃げない(偽の精度を生むので扱わない)。当初は良し悪しを扱わない線だったが、観点を型にできる知見が溜まったので拡張した(下記「本番の実運用セッションを見る場合」)。

検証の構造 (全部これの同型)

状態を仕込む → 改修者が Claude Code を起動して走らせる → transcript を解析して発火・振る舞いを観測 → 後片付け

仕込む状態の大きさが検証の目的で変わるだけ。skill の連鎖発火だけならプローブ一式、実フェーズの振る舞いなら要件定義・基本設計 md を置いた途中状態を仕込む。

(上は sandbox で自分が走らせる場合。既に走った 本番の実運用セッション を分析する場合は「状態を仕込む・走らせる」が無く、下記に従う。)

本番の実運用セッションを見る場合 (sandbox でなく、既に走った実運用)

sandbox 検証は「改修した skill が発火・動作するか」で 的が既知(何を確認するかは改修が決める)。一方、本番の実運用セッション(実際のユーザーの自走など)を解析するときは、的が開いている(何を見るべきか自体が定まっていない)。だから最初に 的=理想を立てる のが最重要ステップになる。仕込み・起動は無く、既に走った transcript を解析するだけ。

理想を先に確定する (最重要・ここを飛ばさない)

  • まず分析対象のフェーズ / 範囲を絞る: 実運用セッションは要件定義 → 基本設計 → 実装と複数フェーズ(=複数 skill)をまたぐのが普通。全部を一度に見ようとせず、見たいフェーズを 1 つ選び、それを統べる skill に絞る。
  • 理想は skill の本文から立てる: 対象の実走を統べる skill(実装フェーズなら slice-tdd、E2E なら e2e 等)の 実際の記述を読み、そこから「あるべき振る舞い」を導く。記憶・想像で立てない(skill の文言と食い違い、照合が的外れになる)。
  • 照らす版を決める: skill は改修で動くので、対象セッションが実際に発火させた版(transcript に残る本文)に照らす。ファイル側の現在版で照らすと、実走時に無かった規律で判定してしまう。
  • 依頼者と合意して固める: 理想の項目をまず短く並べ(発散)、1 項目ずつ深掘りして合意する(収束)。一人で完成形を出さない。認識がズレたまま進めない。
  • 理想が固まる前に照合・判定に進まない: 的が定まる前に実走を時系列で並べても、見づらいだけで分析にならない。理想がズレると照合が全部ズレて手戻りになる。理想の各項目が「何を見るか」を決める。

理想を lens に照合する

  • 確定した理想の各項目を「見る対象」にして、実走が できている / できていない を、証拠(記録上の実際の記述)付き で照らす。
  • 事実(記録から取れるもの)と仮説(内部状態の推測)を分ける。自己分析・因果は捏造されやすいので、判断に効く所は生 transcript で裏を取る(下記手順 4 の「集計を真と扱わない」と同じ)。

Read the full file on GitHub · 246 lines

Files

What ships with it

60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 7d ago First seen · 246 lines · 185 tokens per session scan A 5e6889417856

Subscribe to this mod's changes

harness-verification is a skill published in the GitHub repository sei-newbear/xp-harness (9 stars, last pushed 1mo ago), licensed MIT. It adds 185 tokens to every session and 11,342 once invoked, about $0.0009 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

research-engineer

An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.

davila7/claude-code-templates · 43 tokens

tika-eval-compare

Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".

apache/tika · 50 tokens

neuron-evaluation-engineer

Create and run AI evaluations with datasets, assertions, and output drivers in Neuron AI. Use this skill whenever the user mentions evaluation, testing AI systems, creating evaluators, dataset-driven testing, assertion-based validation, or wants to measure AI system performance. Also trigger for tasks involving…

neuron-core/neuron-ai · 77 tokens

jetson-validate-image

Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.

NVIDIA/skills · 50 tokens

atmos-validation

Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.

cloudposse/atmos · 31 tokens

skill-benchmark

Benchmark AI skill effectiveness by measuring implementation quality against legacy constraints.

HoangNguyen0403/agent-skills-standard · 16 tokens