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.
git clone --depth 1 https://github.com/hahaxiang27/FlowHarnessWrote 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.
[](https://agentmods.dev/commands/hahaxiang27/flowharness/harness.checkpoint)<a href="https://agentmods.dev/commands/hahaxiang27/flowharness/harness.checkpoint"><img src="https://agentmods.dev/badge/commands/hahaxiang27/flowharness/harness.checkpoint.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00024 | $0.02112 |
| Opus 5 | $0.00012 | $0.01056 |
| Sonnet 5 | $0.00005 | $0.00422 |
| Haiku 4.5 | $0.00002 | $0.00211 |
Grade A, and why
harness-checkpoint 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.
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.
How it starts
The opening of the file, as written. The whole thing — 148 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Harness Sprint Checkpoint(Constitution 审查)
上下文管理: 🔄 清空上下文 — 使用子代理执行,确保无偏见的全面审查
指令
对当前 Sprint 的所有代码进行 Constitution 全面合规审查。通常在 Sprint 最后一天执行。
⚠️ 实物验收前置条件(Sprint 47 血教训 · 2026-04-24 · 不可协商)
Checkpoint 不能只看 progress.md 数字。主审(你)在委派子代理审查之前必须完成以下准备:
- 查 Sprint feature 是否涉真实外部服务 / 用户 UI(读 spec / progress 文件)
- 若涉真实外部服务:确认项目凭据可加载,并执行
[REAL_SERVICE_CHECK]至少 1 次成功走查 · 拿到 runId + success 状态证据再委派子代理 - 若涉用户 UI:主审执行
[APP_START_COMMAND]/[UI_START_COMMAND]并启动必要依赖服务 · 浏览器肉眼打开 feature 关联页面 · 截图 ≥ 2 张存证至[E2E_SCREENSHOT_DIR]或/tmp/sprint-{N}-checkpoint/ - 实物证据给子代理:委派任务时必须把 runId / 截图路径 / 接口 response 片段作为证据传入 · 让子代理基于实测证据审查(不是 progress.md 数字)
若主审跳过以上准备直接委派子代理 → 子代理必须显式声明"未收到实物证据" · 对原则 VIII 至少扣 1 CONDITIONAL · 不允许默认 PASS。
输入参数
$ARGUMENTS — 可选,Sprint 编号。不指定则自动检测当前 Sprint。
执行步骤
使用 Agent 工具启动子代理,传入以下任务:
你是 Harness Constitution 审查员。你的职责是对代码进行独立的合规审查,不受之前开发上下文影响。
1. 读取 `.specify/memory/constitution.md`,提取全部7条原则及其 MUST 条款
2. 读取 `.harness/sprints/sprint-{N}-progress.md`,了解本 Sprint 完成的任务和文件变更
3. 逐条审查:
### 原则 I: 架构边界
- 检查服务端、用户界面和其他运行时边界是否符合 Constitution
- 检查是否存在跨边界直接耦合的实现
- 检查所有跨边界通信是否通过项目定义接口
### 原则 II: 接口契约驱动
- 读取 `specs/[FEATURE_ID]/contracts/` 或 `[API_CONTRACT_ROOT]` 中所有已实现的接口契约
- 逐个端点对照实现,检查路径/方法/参数/响应是否一致
- 检查是否存在契约中未定义的接口
### 原则 III: 测试纪律
- 检查认证、权限和高风险业务规则是否达到要求的分支覆盖
- 检查已实现接口是否有契约测试
- 检查测试是否可在 CI 环境独立运行(无外部依赖)
### 原则 IV: 外部服务抽象层
- 检查 `[EXTERNAL_SERVICE_NAME]` 调用是否通过 `[EXTERNAL_SERVICE_ADAPTER]` 接口
- 检查 Prompt/配置是否外置在 `[PROMPT_ROOT]` 或项目约定位置中
- 检查是否配置了超时和降级策略
### 原则 V: 可观测性
- 检查接口请求日志是否有结构化记录
- 检查外部服务调用是否有指标记录
- 检查错误是否有唯一追踪 ID
### 原则 VI: 简单优先
- 检查是否引入了不必要的抽象层
- 检查新依赖是否在 Complexity Tracking 中有记录
- 检查是否引入了不必要的架构拆分
### 原则 VII: 安全与权限
- 检查所有接口是否有认证检查
- 检查敏感操作是否有审计日志
- 检查是否存储明文密码或暴露内部异常
### 原则 VIII: 用户界面与真实服务验收(Sprint 47 沉淀 · 实物验收硬约束)
- **涉真实外部服务场景**:主审给的 runId 是否 success · `[EXTERNAL_SERVICE_METRIC]` 非零 · failed 为 0
- **涉用户 UI**:主审给的截图是否覆盖 feature 关键页面 ≥ 2 张 · 截图里关键字段非空
- **人工验证节点**:planner 规划时 identify 的 HV 节点是否全部有签收记录(progress.md 里 `👁 HV-M PASS`)
- **Spec 边界承诺**:是否存在"零 X 改动"类声明被本 Sprint 打破却未扣分(scope 漏洞诚实承认)
- 若主审未给实物证据(runId / 截图 / HV 签收),本原则**自动扣至少 1 CONDITIONAL**,不默认 PASS
- `[E2E_TOOL]` 的 Mock 模式通过 ≠ PASS(只能证明主路径不崩)
### 原则 IX: 用户界面视觉质量
- 布局/间距/色板合规 · HV 节点截图视觉审查
### 原则 X: Corrector 纪律
- Sprint 内 Corrector 轮次计数(含批次内收敛)· 非零不等于 FAIL
- 每轮 Corrector 是否严格走 `/harness.fix`(先审 spec 定预期 · TDD Red/Green · 回归 · 实物验收)
- 未经 /harness.fix 的乱改代码 → FAIL · 合规则按轮次扣:0 轮满分 · 1-2 轮扣 1-2 · 3+ 轮 FAIL
输出合规审查报告:
| 原则 | 状态 | 检查项数 | 通过数 | 问题详情 / 扣分依据 |
|------|------|---------|--------|---------|
### 得分规则(诚实扣分 · 避免 100/100 凑分)
- 全部核心原则 × 10 为满分
- 每原则按扣分项逐条扣(1 扣 1 · 2 扣 2 · 硬违反至少扣 5)
- 得分率 ≥ 99% PASS · ≥ 95% CONDITIONAL PASS · < 95% FAIL
- Sprint 47 教训:诚实扣分好过假 100/100 · scope 漏洞 / DEFERRED 未跑 / Corrector 发生都应扣
### Checkpoint 演进轨迹(必须)
若本 Sprint 的 Checkpoint 经历过撤回 / 重跑 · 必须在 checkpoint.md 显式记录:
- 初判(日期 + 得分 + 审查方法)
- 撤回原因(具体 SEV + 违反原则)
- 重跑(基于实测证据 + 对比初判差异)
- 纪律沉淀(教训去向 · memory / template)
总体结论:PASS / CONDITIONAL PASS / FAIL(基于得分阈值)
如有违反,输出具体的文件路径和行号,以及修复建议。
更新 sprint-*-progress.md 中的 Constitution 审查结果。
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.
- 7d ago First seen · 148 lines · 24 tokens per session scan A 3d7886395f98
harness-checkpoint is a command published in the GitHub repository hahaxiang27/FlowHarness (4 stars, last pushed 2mo ago), licensed MIT. It adds 24 tokens to every session and 2,112 once invoked, about $0.0001 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.
Other commands, from other repositories
gh-triage
GitHub OSS maintainer lifecycle triage, review, and approval management.
sync-pr-dashboard
Sync the NI&D GitHub Project board — populate PR authors, sync reviewers to assignees, add shared repo PRs, classify areas.
rfc
RFC process for cross-team decisions. Create, track, and close RFCs. Accepted RFCs auto-create ADRs.
discover
Run a full product discovery cycle — from outcome definition through opportunity mapping, prioritisation, and experiment design. Use when the team isn't sure what to build next, or before writing a PRD for a complex feature space.
migrate
Migrate existing PROJECT.md to the latest greatcto schema — appends missing fields without touching existing values.
exception
Signed gate-exception registry — replace ad-hoc --admin / --no-verify bypasses with an auditable, signed, expiring record (who · why · gate · scope · expiry). Create / list / check exceptions.