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.
npx skills add YuDefine/nuxt-supabase-starter --skill verificationgit clone --depth 1 https://github.com/YuDefine/nuxt-supabase-starterWrote 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/skills/yudefine/nuxt-supabase-starter/verification)<a href="https://agentmods.dev/skills/yudefine/nuxt-supabase-starter/verification"><img src="https://agentmods.dev/badge/skills/yudefine/nuxt-supabase-starter/verification/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/yudefine/nuxt-supabase-starter/verification"><img src="https://agentmods.dev/badge/skills/yudefine/nuxt-supabase-starter/verification.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00016 | $0.01010 |
| Opus 5.5 | $0.00006 | $0.00404 |
| Sonnet 5.5 | $0.00003 | $0.00202 |
| Haiku 4.5 | $0.00002 | $0.00101 |
Grade A, and why
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 8d 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verification infrastructure
verification 是 consumer verification infrastructure 的統一入口：create 建立 verify-<app> skill 與 feature map 並實證一次閉環，maintain 對帳既有 feature map 並 live re-prove；只維護 verification infrastructure，不宣告產品驗收。
改編自 Lauren Tan 的 pstack create-verification-skill／maintain-verification-skill，授權見 references/UPSTREAM.md。
SOP
Phase 1 -- 判定 mode 與定位 target
- READ 找
<skills-root>/verify-*/SKILL.md（<skills-root>依 runtime：Claude.claude/skills/、Codex.agents/skills/；找不到該目錄就回報標準未送達，不猜一個）。 - THINK 零個 → create（Phase 2）；一個且要更新或重跑 → maintain（Phase 5）；多個 → 請使用者指定，不靠名稱相似度猜。只收集 UI 截圖交
/reviewscreenshot；implementation 與規格對帳讀specs/truth/**與該 work 的 carrier 自行對帳，動到規格回交 truth owner skill。
Phase 2 -- create：訪談 repo
- THINK 先讀取
rules/repo訪談六欄判準.md，從 repo 與實跑結果回答 Surface／Launch／Doctor／Drive／Observe／Isolate；不能 build 或 launch 就停下回報 baseline blocker。
Phase 3 -- create：產生 verify skill 與 feature map
- WRITE 先讀取
rules/verify-skill產出判準.md、templates/verify-app.md與templates/verify-app.example.md，在 canonical 位置寫<skills-root>/verify-<app>/SKILL.md。 - WRITE 先讀取
rules/feature-map結構判準.md、templates/verification-map-README.md、templates/verification-map-README.example.md、templates/verification-feature.md與templates/verification-feature.example.md，寫features/README.md與最重要的 3–5 個 feature。
Phase 4 -- create：驗結構並實跑一次閉環
- DELEGATE 執行
node <skills-root>/verification/scripts/check-feature-map.mjs <skills-root>/verify-<app>，exit 0 才往下；結構紅時只修 verification skill，不改產品 code 來迎合文件。 - DELEGATE 實跑一次完整閉環 Launch → Doctor → Drive one mapped feature → Capture evidence → Cleanup；失敗 iteration 也 cleanup，最後確認 evidence 在 cleanup 後仍存在。
- WRITE 先讀取
rules/create完成證據判準.md，逐格勾完成證據，交代 target、已實跑 feature、evidence path、unreachable prerequisites，並指出後續用/verification maintain。
Phase 5 -- maintain：驗索引與 source wave
- DELEGATE 執行同一支
node <skills-root>/verification/scripts/check-feature-map.mjs <skills-root>/verify-<app>，修 missing／extra／duplicate／dead index entry；同時先讀取rules/維護編輯範圍判準.md確認可改範圍。 - DELEGATE 先讀取
rules/source-wave與對帳判準.md，依 clade routing threshold 以 read-heavy prescan 收集每個 feature 的 source evidence，再依規則對帳。
What ships with it
18 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.
- references/UPSTREAM.md 1.9 KB
- rules/create完成證據判準.md 2.1 KB
- rules/feature-map結構判準.md 3.1 KB
- rules/live-pass紀律判準.md 1.6 KB
- rules/repo訪談六欄判準.md 1.9 KB
- rules/source-wave與對帳判準.md 1.6 KB
- rules/verify-skill產出判準.md 2.8 KB
- rules/維護outcome判準.md 2.8 KB
- rules/維護編輯範圍判準.md 1.7 KB
- scripts/check-feature-map.mjs 3.8 KB runs code
- templates/verification-feature.example.md 643 B
- templates/verification-feature.md 333 B
- templates/verification-maintenance-report.example.json 483 B
- templates/verification-maintenance-report.json 578 B
- templates/verification-map-README.example.md 1.3 KB
- templates/verification-map-README.md 1.1 KB
- templates/verify-app.example.md 1.6 KB
- templates/verify-app.md 1.0 KB
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.
- 8d ago Changed · -1 lines be1aab8d6f0e
- 9d ago Changed · +20 lines eb59ba480582
- 23d ago Changed 2ed89b302d82
- 24d ago Changed · +1 lines cd12812aad2a
- 1mo ago First seen · 29 lines · 16 tokens per session scan A 0342e55b9fc5
verification is a skill published in the GitHub repository YuDefine/nuxt-supabase-starter (44 stars, last pushed 2d ago), licensed MIT. It adds 16 tokens to every session and 1,010 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-09-14.
Other skills, from other repositories
prd-v07-implementation-loop
Execute implementation within EPICs following test-first development, continuous SoT updates, and code traceability during PRD v0.7 Build Execution. Triggers on requests to start building, implement an epic, begin coding, or when user asks "start building", "implement epic", "coding", "development", "build execution"…
prd-v07-test-planning
Define test cases BEFORE implementation, ensuring every API, business rule, and user journey has verifiable acceptance criteria during PRD v0.7 Build Execution. Triggers on requests to define tests, plan test coverage, create test cases, or when user asks "define tests", "test planning", "what to test?", "test cases"…
conductor-implement
Execute tasks from a track's implementation plan following TDD workflow.
python
Python development with ruff, mypy, pytest - TDD and type safety.
etl-stabilization
Phase-based orchestrator for fixing SnowConvert ETL conversion gaps in ETL-to-Snowflake migrations (SSIS, Informatica, and other platforms). Analyzes units holistically, creates a ROADMAP with test strategies, and executes fixes through phased TDD with cross-phase learning. Use when user says "fix ETL unit", "fix SSIS…
bug-fix
Regression-first TDD bug fix workflow with real tests. Use when: fixing bugs, resolving regressions, addressing runtime errors, debugging failures, or when the user reports something broken. Triggers on words like: bug, broken, error, crash, regression, not working, fix.