verify-gate
Skill4 starsChanged 54 days agoMITView on GitHub
关项前按档位只核核心证据并写验证报告；证据深度随档位渐进，不逐步全验。
7 other projects use a version of this.
Adopted by
Most recently taken up by
- kuhlman-labs/fishhawktheir own version, first seen Oct 31, 2026
- Ces107/agent-forgetheir own version, first seen Oct 25, 2026
- Jaxsbr/sabstheir own version, first seen Oct 20, 2026
- feelthefusion/skill-starter-kittheir own version, first seen Oct 13, 2026
- svy04/ballasttheir own version, first seen Aug 23, 2026
- piprail/piprailtheir own version, first seen Jun 10, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- pskoett/pskoett-ai-skills★ 327same name
- svy04/ballast★ 73same name
- kuhlman-labs/fishhawk★ 9same name
- piprail/piprail★ 9same name
- Ces107/agent-forge★ 0same name
- feelthefusion/skill-starter-kit★ 0same name
- Jaxsbr/sabs★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository112 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .codex/AGENTS.mdAGENTS.md
- .kimi-code/AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/全局.mdcCursor rule
- anim-event-hookSkill
- anim-set-checklistSkill
- anti-cheat-hook-checkSkill
- 产物登记Skill
- assemble-craft-flowSkill
- attr-family-syncSkill
- attribute-frameworkSkill
- audio-fmod-checklistSkill
- auto-test-scaffoldSkill
- bind-rig-checklistSkill
- blockout-passSkill
- bug-report-writeSkill
- build-acceptanceSkill
- build-gate-checklistSkill
- case-automation-mapSkill
- character-asset-checklistSkill
- ci-smokeSkill
- client-bugfixSkill
- client-combat-frame-debugSkill
- collab-protocolSkill
- combat-feel-checklistSkill
- combat-flow-designSkill
- combat-modelingSkill
- compat-smokeSkill
- concept-key-artSkill
- copy-passSkill
- counter-matrix-passSkill
- damage-formula-passSkill
- data-readiness-checkSkill
- deliverable-sheetsSkill
- dependency-mapSkill
- device-matrix-passSkill
- diagram-packSkill
- dialogue-passSkill
- doctorSkill
- economy-loop-analysisSkill
- encounter-scriptSkill
- env-asset-checklistSkill
- event-specSkill
- excel-com-writeSkill
- excel-formatSkill
- excel-readSkill
- experience-critiqueSkill
- export-naming-gateSkill
- export-pipeline-fixSkill
- feature-gdd-sliceSkill
- feature-vertical-sliceSkill
- file-pack-layoutSkill
- fmod-bank-buildSkill
- 交接Skill
- iap-catalog-checkSkill
- import-validateSkill
- inflation-stressSkill
- The first 60 shown.
The file
Files
验证
第一层关项只认报告 verdict 为 pass。「真通过」= 本档位该核的核心证据已打开且相符。
验证报告是关项级一份。中间步骤靠制作自检。
| 档位 | 验证种类 | 核什么 | 大约证据条数 |
| 讨论 | none | 无交付则不写报告 | 0 |
| 短活 | smoke | 主产物在、非空、抽一眼内容对 | 1 |
| 已知配方 | schema 或 smoke | 主产物 + 关键约束 | 1～2 |
| 多职种 | playtest 或 schema | 上栏 + 跨职种接口 | 2～3 |
| 硬闸 | build / release | 上栏 + 构建或发版输出 | 3～5 |
怎么做
1. 定核什么
- 读加载计划：
bead_id、tier、verify_kind、主交付路径。 - 只列本档位核心项；每项一个已存在路径。
verify_kind=none且无交付：停，不写报告。
做到：1～5 条核心项。
2. 打开证据
- 逐项打开：在、非空、与预期相符。
- 改表核回读或 diff；构建看日志或产物路径。
- 打不开或内容不符 → 本项失败，不凑绿。
做到：已打开路径列表。
3. 写报告并登记
路径：.harness/artifacts/<工作项>/verify-report.json
| 字段 | 写什么 |
bead_id | 工作项编号 |
verify_kind | 与计划一致（不含 none） |
command | 本档位核对动作一句 |
exit_code | 全过 0，失败 1 |
evidence_paths | 已打开的核心路径 |
verdict | pass / fail |
notes | 失败时写哪条不过 |
核心项全过且路径非空 → pass。
随后追加一行到 .harness/artifacts/index.jsonl（ts、bead_id、skill=verify-gate、path、summary）。
再写 .harness/state.json 的 progress，并往 .harness/memory/tasks.jsonl 追加一行。短活只登记报告自身。多文件批次或换手再点 artifacts-append。
做到：报告字段齐全；索引尾部可见本行。
4. 关项
仅 pass 后关项。fail 按 notes 回流制作。
完成检查
- 档位与
verify_kind对齐 - 证据已打开并写入报告
- 报告已登记进索引
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.