Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
Offline receipt verifier. Ed25519 signature verification without contacting any server.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"veritasacta-verify": {
"command": "npx",
"args": [
"-y",
"@veritasacta/verify"
]
}
}
}在终端中运行这条命令
claude mcp add veritasacta-verify -- npx -y @veritasacta/verify粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"veritasacta-verify": {
"command": "npx",
"args": [
"-y",
"@veritasacta/verify"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"veritasacta-verify": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@veritasacta/verify"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月9日
发布者(tomjwxf)不是仓库所有者(VeritasActa);可能是其他人代为上架的
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 0 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(0.11.1)于15天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:Spec Driven Development, Remote Desktop Commander, SwarmMemo
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Talk with agent swarms: rooms, private messages and paid tasks for AI agents
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Check signed records yourself, without contacting the issuing service.
An offline CLI and a shared browser receipt engine. Supply a receipt and a public key obtained through a separate trusted channel, then apply your own trust policy. No account, proxy, signing daemon or onboarding step is required to verify a receipt.
Try the browser sample · Receipt specification · Conformance evidence
npm install -g @veritasacta/verify@0.11.1
verify receipt.json --key YOUR_INDEPENDENTLY_OBTAINED_PUBLIC_KEY
The key is 64 hexadecimal characters. Do not treat a key supplied with an unknown receipt as an independent trust pin. This verification path runs offline once the package is installed; installation and explicitly requested URL/JWKS lookups use the network.
An Acta receipt or audit bundle is never checked with a key taken from the file being checked. A receipt whose signed bytes carry a key (public_key, verification_key, verification_jwk) other than the one you supply fails with key_mismatch, and an envelope whose payload names an issuer_id other than its signature's kid fails with issuer_kid_mismatch. An audit bundle's own verification.signing_keys are used only if you pass --trust-bundle-keys, and a valid result then shows only that the receipts match keys in the same file. Gate records, macro track records, governed receipts, proof packs, context packs, repository evidence, standard files and run manifests carry their own key; checked without --key, the result is undecidable (exit 2). --allow-carried-key accepts it (exit 0) and says the signature proves nothing about who signed; --strict refuses it even then. Key windows apply to the issued_at the signature covers, and disclosures are checked against the commitments the signature covers; a disclosure that cannot be applied is not checked (exit 2).
To try the public example, download this receipt and use its demonstration-only key:
verify current-receipt.json --key ebb8cf3efc78717c50e7c438c0ad5498f8fa4e09607fea1e030ac129b5965225
A valid signature is not proof of claim truth, complete action coverage, signer authority or recipient acceptance. A signed timestamp is an issuer assertion, not an independent timestamp.
| Outcome | Meaning | Next step |
|---|---|---|
| Verified / exit 0 | The checks implemented for this profile passed | Apply identity, authority and reliance policies; read warnings and binding scope |
| Failed / exit 1 | A verification check failed | Check the expected key, signature, structure and reported error |
| Not checked / exit 2 | The verifier could not decide | Supply missing material or use a supported format/algorithm |
Exit 1 does not prove tampering. Supplying the wrong expected key can produce the same result. Some legacy JSON error categories use the name tampered for compatibility; it denotes failed verification, not a forensic conclusion. See error codes and the profile-specific output.
Both use src/detect.js, src/engines/ed25519-receipt.js and the same canonicalizer. The browser bundle is published in browser/, with a SHA-256 build manifest.
The browser accepts individual Ed25519 envelopes, v2 structured receipts, v1 flat receipts and historical website receipt_version: "1.0" receipts. That historical profile excluded public_key from signing and is explicitly labelled; it is not silently promoted to draft-03 conformance. All browser profiles require an external key. Unsupported formats are not checked, not reported as bad signatures.
Both entry points reject duplicate JSON members before parsing. The browser limits input to 1 MiB and depth 64; the CLI file parser allows 50 MiB and depth 128. Bundles and specialist profiles, including selective disclosure, VOPRF, Knowledge Units, Gate evidence and governed-run manifests, require the CLI. Profile success only establishes the checks named in that output, not universal completeness.
| Task | Guide |
|---|---|
| Verify bundles, chains, prompts or specialist artifacts | Advanced reference |
| See this CLI's results on TRACE's Acta fixtures | TRACE fixture results |
| Issue receipts or configure a proxy/daemon | Issuance and adapters |
| Understand threats and limitations | Threat model |
| Report a vulnerability | Security policy |
| Inspect exact release changes | Changelog |
| Run the tests or contribute a fixture | Contributing |
| Understand project decisions and stewardship | Governance |
verify --self-check recomputes monitored source hashes against a bundled integrity commitment (the Sigil). It detects drift from that commitment. Someone replacing both the code and commitment can create a new internally consistent package.
It does not independently authenticate the publisher or establish a canonical release. Pin a reference obtained independently and inspect release provenance. --attest is a local, self-signed integrity statement; separately pin its key before relying on it.
Verifier: [0.11.1](https://github.com/VeritasActa/verify/releases/tag/v0.11.1). Status reviewed 2026-10-25.
Internet-Drafts are proposals, not adopted IETF standards or endorsements.
release-manifest.json is the source for version/status blocks in this README, the website, the organisation profile and draft index. node scripts/sync-public-metadata.mjs --check checks sibling repository copies. No draft is revised or resubmitted by this synchronization.
Maintained by Tom Farley / ScopeBlind. Independent verification does not imply independent organizational governance. Contributions should include reproducible fixtures and negative cases. The contribution guide covers a clean checkout, package checks and useful first contributions. See the public project evidence for reproducible interoperability results and their limits.
npm ci
npm test npm run check:browser npm run check:sigil npm run test:conformance
Apache-2.0. See LICENSE and PATENTS.md for the actual terms. The Acta public record is a separate reference application. ScopeBlind's commercial applications are not prerequisites for verification.