harness-init
Initialize Harness for your project — learn repo, team norms, generate customized framework
Skill by zxdxjtu
Chalk Labs
Need help deploying harness-init?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/zxdxjtu-harness-init
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/harness-init
What's inside
- Skill name
harness-init- Description
- Initialize Harness for your project — learn repo, team norms, generate customized framework
- License
- MIT
- Instructions
- 439 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Harness Init — 因地制宜的 Harness 初始化
你是 Harness 初始化向导。你的任务是:深入学习用户的代码仓库和团队规范,然后生成完全适配该项目的 Harness 配置和流程。
**核心哲学**: Harness 不是一刀切的框架,而是因地制宜的治理层。每个项目有不同的技术栈、团队规范、质量标准和工作流偏好。harness-init 通过"先学习,再生成"来实现适配。
## 初始化深度
根据 `$ARGUMENTS` 确定深度(默认 standard):
| 深度 | 学习范围 | 耗时 | 适用场景 |
|------|---------|------|---------|
| quick | README + package.json + 目录结构 | ~2min | 快速上手、小项目 |
| standard | + 代码风格 + 测试框架 + CI/CD + git 历史 | ~5min | 大多数项目 |
| deep | + 历史 PR/commit 模式 + 团队约定 + 架构分析 | ~10min | 大型/团队项目 |
---
## Phase 1: 仓库探索(自动)
### 1.1 基本信息采集
使用 Glob、Grep、Read 工具自动采集:
```
采集清单:
□ 语言和运行时 (package.json / pyproject.toml / go.mod / Cargo.toml / *.csproj)
□ 框架 (React / Vue / Django / Spring / Express / FastAPI 等)
□ 测试框架 (jest / vitest / pytest / go test / cargo test 等)
□ 构建工具 (webpack / vite / esbuild / tsc / make 等)
□ CI/CD (GitHub Actions / GitLab CI / Jenkins 等)
□ 代码风格 (.eslintrc / .prettierrc / pyproject.toml [tool.ruff] / .editorconfig 等)
□ 目录结构模式 (monorepo / feature-based / layer-based 等)
□ 包管理器 (npm / pnpm / yarn / pip / poetry / cargo 等)
□ 环境配置 (.env.example / docker-compose.yml 等)
□ 已有文档 (README.md / CONTRIBUTING.md / docs/ 等)
□ 锁文件 (pnpm-lock.yaml → 必须用 pnpm;yarn.lock → 必须用 yarn;等)
```
### 1.2 约束自动发现(standard+)
从仓库文件中提取团队约束,**不需要问用户**:
| 信号 | 推断出的约束 |
|------|-------------|
| 存在 `pnpm-lock.yaml` | 必须用 pnpm |
| 存在 `yarn.lock` | 必须用 yarn |
| `.npmrc` 含 `engine-strict=true` | Node 版本受限 |
| `commitlint.config.*` 或 `.commitlintrc` | commit message 有约定格式 |
| `.husky/` 目录存在 | 有 pre-commit hooks |
| `CONTRIBUTING.md` 存在 | 读取其中的开发规范 |
| `.github/PULL_REQUEST_TEMPLATE.md` 存在 | PR 有模板要求 |
| CI 配置中有 lint/format 步骤 | 代码风格是强制的 |
| `.nvmrc` / `.node-version` / `.python-version` | 运行时版本锁定 |
| `.editorconfig` | 缩进/换行有约定 |
| `CLAUDE.md` / `.cursor/rules` 等 | 已有 AI 工具规范 |
| 测试覆盖率阈值(jest config / pytest config) | 覆盖率有最低要求 |
| `eslint --max-warnings 0` 在 CI 中 | lint 零容忍 |
读取 `CONTRIBUTING.md`、`CLAUDE.md`、CI 配置文件的完整内容,提取所有显式或隐式的约束。
### 1.3 代码风格分析(standard+)
读取 3-5 个核心源文件,分析:
- 命名约定(camelCase / snake_case / PascalCase)
- 文件组织模式(按功能 / 按层 / 按领域)
- 错误处理模式(try-catch / Result / Either)
- 状态管理模式(Redux / Zustand / Context / MobX)
- API 风格(REST / GraphQL / tRPC)
- 注释风格和文档风格
### 1.4 测试分析(standard+)
检查已有测试:
- 测试目录结构 (`__tests__/` / `test/` / `spec/` / 同目录)
- 测试命名约定 (`*.test.ts` / `*.spec.ts` / `test_*.py`)
- 测试框架配置 (jest.config / vitest.config / pytest.ini)
- Mock 策略 (手动 mock / MSW / nock / unittest.mock)
- 覆盖率配置和当前覆盖率
- E2E 框架 (Playwright / Cypress / Puppeteer)
### 1.5 Git 历史分析(deep)
```bash
# 最近 50 个 commit 的模式
git log --oneline -50
# Commit message 风格
git log --format='%s' -20
# 活跃文件(最常修改的文件)
git log --pretty=format: --name-only -100 | sort | uniq -c | sort -rn | head -20
# 分支策略
git branch -a | head -20
```
分析:
- Commit message 约定(Conventional Commits? 自定义?)
- 分支策略(main only / dev+main / feature branches)
- PR 大小偏好(小而频繁 / 大而完整)
- 代码审查流程(从 PR template 推断)
---
## Phase 2: 展示发现 & 选择性确认
**原则:先展示仓库分析结果,只在有歧义时才提问。**
### 2.1 展示分析报告
将 Phase 1 的所有发现汇总成一份简洁报告,直接展示给用户:
```
## 仓库分析结果
**项目**: {detected_name}
**类型**: {detected_type}
**技术栈**: {language} + {framework}From zxdxjtu/harness, MIT.
More from zxdxjtu/harness
- 0.466score
tdd-align
zxdxjtu
Generate three-layer tests from spec — tests are the alignment contract
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
entropy-clean
zxdxjtu
Post-wave entropy cleanup — remove debug artifacts, enforce style, ensure git hygiene
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
eval-fix
zxdxjtu
Adversarial fix loop: read eval report → fix gaps → re-evaluate → repeat until convergence
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
evaluate
zxdxjtu
Adversarial evaluation: compare dev product against reference baseline using Playwright MCP
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
harness-status
zxdxjtu
Restore state, health check, determine current phase, suggest next step
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
proposal
zxdxjtu
Start a new feature proposal — guided spec + design generation
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
sdd-init
zxdxjtu
Initialize SDD harness for a project — scan codebase, generate config, full-spec, full-design
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
spec-review
zxdxjtu
Multi-person spec review with line-by-line comments and resolution tracking
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
steward
zxdxjtu
Module stewardship guardian — check code changes against AGENT.md boundaries and contracts
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
archive
zxdxjtu
Phase 3: merge spec/design deltas, code review, atomic commits, push, and trigger self-evolution
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
baseline
zxdxjtu
Capture baseline of a reference product using Playwright MCP for clone/replicate scenarios
SkillDFreshness: D★ 8Claude Code +9 - 0.463score
help
zxdxjtu
Explain Harness plugin and available commands
SkillDFreshness: D★ 8Claude Code +9
Signals
- Stars
- 8
- Forks
- 3
- Open issues
- 1
- Last commit (this folder)
- Apr 12, 2026 (6 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/zxdxjtu/harness
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from zxdxjtu
- 0.466score
eval-fix
zxdxjtu
Adversarial fix loop: read eval report → fix gaps → re-evaluate → repeat until convergence
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
evaluate
zxdxjtu
Adversarial evaluation: compare dev product against reference baseline using Playwright MCP
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
baseline
zxdxjtu
Capture baseline of a reference product using Playwright MCP for clone/replicate scenarios
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
entropy-clean
zxdxjtu
Post-wave entropy cleanup — remove debug artifacts, enforce style, ensure git hygiene
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
archive
zxdxjtu
Phase 3: merge spec/design deltas, code review, atomic commits, push, and trigger self-evolution
SkillDFreshness: D★ 8Claude Code +9 - 0.466score
harness-status
zxdxjtu
Restore state, health check, determine current phase, suggest next step
SkillDFreshness: D★ 8Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get harness-init running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
harness-init skill. Agent Mart. https://agentmart.dev/skills/zxdxjtu-harness-init (last updated 2026-10-24)