原型 · 预览版:收录官方注册中心里的全部服务器,每 30 分钟同步一次。评价和账户功能将在后续阶段推出。
PopNx MCP为使用 AI 的人准备的 MCP 服务器 · 用大白话讲清楚

Veilfile

作者:veilfile版本 1.0.0i★ –i在云端运行
注册名称io.github.veilfile/veilfile

安装

概览

作者提供的说明(英文原文)

Private, expiring artifact hosting for AI agents. Upload via API or MCP; links die on their own.

PopNx MCP 的三大支柱

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。

它能帮你做什么

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。

  • 连接方式: 在云端运行
  • 传输方式: Streamable HTTP
  • 分类: 其他
  • 你需要填写:Authorization

安装

选择你使用的 AI 应用,然后复制设置。

安装之前

标为机密的值是你账户的钥匙。千万不要分享给别人,也不要粘贴到聊天里。

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 …”。

这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。

在终端中运行这条命令

claude mcp add --transport http veilfile https://veilfile.com/mcp/ --header 'Authorization: your-value'

运行前请先填写空着的值。

在 Cursor 中打开

或者自己复制设置

粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json

{
  "mcpServers": {
    "veilfile": {
      "url": "https://veilfile.com/mcp/",
      "headers": {
        "Authorization": "your-value"
      }
    }
  }
}

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。

在 VS Code 中打开

或者自己复制设置

粘贴到项目中的 .vscode/mcp.json

{
  "servers": {
    "veilfile": {
      "type": "http",
      "url": "https://veilfile.com/mcp/",
      "headers": {
        "Authorization": "your-value"
      }
    }
  }
}

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。

这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。

你的选择通过 Cookie 保存在这台设备上,无需账户。

需要你填写的值

  • Authorization 机密 必填 HTTP header
    Bearer API key, issued in the Veilfile dashboard after email verification. Shown once; store as an env var.

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。

它能做什么

我们还没有读取这个服务器的工具列表;我们会轮流检查各个端点。

基础检查i

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6日

  • ✓ 通过

    作者与仓库一致

    注册名称已通过 GitHub 账户 veilfile 验证,该账户拥有此仓库

  • … 尚未检查

    仍在维护

    我们还没有从 GitHub 读取这个仓库

  • … 尚未检查

    有许可证

    我们还没有从 GitHub 读取这个仓库

  • … 尚未检查

    端点有响应

    尚未检查;我们会轮流检查各个端点

  • … 尚未检查

    使用 OAuth 登录

    尚未检查;我们会轮流检查各个端点

  • … 尚未检查

    工具描述中没有隐藏指令

    尚未检查;我们会在检查端点时读取工具列表

与“其他”中的其他服务器比较i

  • 星标数多于此分类中其他 32417 个服务器的 0%
  • 发布到注册中心的时间比此分类中其他 32417 个服务器的 77% 更新
  • 基础检查结果等于或优于此分类中其他 32417 个服务器的 100%

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。

常见问题

Veilfile 安全吗?

它在 1 项适用的基础检查中通过了 1 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。

它需要哪些密钥或代码?

你需要填写:Authorization

它在我的电脑上运行还是在云端?

它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。

它还在维护吗?

GitHub 仓库最后更新于–,最新版本(1.0.0)于10天前进入注册中心。

哪些应用可以使用它?

支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。

有什么可以替代它?

同一分类中有:Spec Driven Development, Remote Desktop Commander, FOGA node

替代选择

同一分类中其他作者的服务器,按星标数排序。

如何看卡片:★ 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

Spec Driven Development

作者:formulahendry

MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.

在你的电脑上运行★ 440MIT14个月前更新
安装

Remote Desktop Commander

作者:desktop-commander

Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.

在云端运行★ 211昨天更新
安装

FOGA node

作者:garyyang-finchip

A Finch OG Agent cell as an MCP server: read the world, dry-run and take ERC-8414 tenders by policy.

电脑或云端今天更新
安装

作者的 README

打开 README

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。

Veilfile — private, ephemeral artifact hosting for AI agents

When a coding agent in CI or a cloud session produces a screenshot, log bundle, or HAR file for review, it needs a private URL to put it at — not a public repo. Veilfile is the sanctioned place: upload via API key (or MCP), get back an unguessable, expiring URL. Uploads are scanned for secret shapes and flagged (blocked only when the caller passes on_secret=reject; file contents are never logged).

  • API: POST /api/v1/artifacts (multipart, Authorization: Bearer vlf_…) → {id, url, expires_at, size_bytes, secret_flags[]}. GET /a/ serves the file (the 256-bit token *is* the auth). List (bare JSON array) and revoke endpoints included.
  • MCP: hosted server at POST /mcp (JSON-RPC over HTTP; tools upload_artifact, create_upload_url, list_artifacts, revoke_artifact), authenticated with a vlf_ key or an OAuth token.
  • CLIs: veilfile on PyPI (cli/) and npm (npm/): login (device flow), logout, status, push, pull.
  • Dashboard: signed-in pages at /app/, Django views on the same shell as the rest of the site: Overview (usage vs plan caps), Files (secret-flag notes
  • revoke), Keys (a new key is shown once), Connected apps and Billing (plan switching and the Stripe billing portal).
  • Landing page: / (public). Agent docs: /llms.txt, docs/API.md, /docs/.
  • Agents working on this repo: read AGENTS.md.

See SPEC.md for the product spec.

Plans

PlanPriceUploads/moMax TTLStorageAPI keys
Free$01007 days1 GB1
Team$4/mo2,00090 days25 GB5
Team Plus$6/mo4,00090 days40 GB10
Scale$12/mo10,000365 days100 GBunlimited
Team (yearly)$40/yr2,00090 days25 GB5
Team Plus (yearly)$60/yr4,00090 days40 GB10
Scale (yearly)$120/yr10,000365 days100 GBunlimited

Yearly billing costs ten months' price, so two months are free. A yearly plan has the same limits as the monthly one.

The source of truth is backend/drop_core/plans.py (PLANS, and ANNUAL_PRICE_USD for the yearly prices; Stripe prices live in Stripe and the STRIPE_PRICE_* variables). backend/drop_core/tests_fix_docs.py fails when a doc or template states a different price or cap.

Quickstart (local dev)

# Backend

python3 -m venv venv && ./venv/bin/pip install -r backend/requirements.txt export DJANGO_DEBUG=True DJANGO_SECRET_KEY=dev-only-key ./venv/bin/python backend/manage.py migrate ./venv/bin/python backend/manage.py createsuperuser # dashboard login ./venv/bin/python backend/manage.py runserver # :8000

Sign up at http://localhost:8000/accounts/signup/ (or log in), open /app/, create an API key, then:

curl -H "Authorization: Bearer vlf_..." \

-F "file=@screenshot.png" \ http://localhost:8000/api/v1/artifacts/

Environment variables

No secrets are committed to this repo. Copy the names below into a local .env (gitignored) or your host's env config. Billing is live in production (sk_live_..., live price IDs, Stripe Managed Payments); use a Stripe sandbox/test key only on a local or staging copy.

VariableRequiredDefaultNotes
DJANGO_SECRET_KEYprod—Required when DJANGO_DEBUG is false
DJANGO_DEBUGnoFalseSet True for local dev and for tests
DJANGO_ALLOWED_HOSTSprod—Comma-separated, e.g. veilfile.com
DJANGO_CSRF_TRUSTED_ORIGINSnoderived from DJANGO_ALLOWED_HOSTSComma-separated origins
DATABASE_URLprodsqlite (dev only)Postgres URL. Production refuses to start without it, or with a sqlite:// URL, unless DJANGO_ALLOW_SQLITE=True. User, password and name are percent-decoded; ?sslmode=... is passed to the driver
DJANGO_ALLOW_SQLITEno—True lets production mode run on SQLite on purpose (local docker run, image build)
DJANGO_CONN_MAX_AGEno60Persistent DB connection lifetime (seconds)
DROP_BASE_URLprodhttp://localhost:8000Public URL; artifact links are built from it
DROP_STORAGEnolocallocal (dev) or s3 (prod)
DROP_LOCAL_DIRnobackend/media/artifactsLocal storage root
AWS_S3_ENDPOINT_URLprod (s3)—R2 endpoint, e.g. https:// .r2.cloudflarestorage.com
AWS_S3_BUCKETprod (s3)—R2 bucket name
AWS_ACCESS_KEY_IDprod (s3)—R2 API token (access key)
AWS_SECRET_ACCESS_KEYprod (s3)—R2 API token (secret)
STRIPE_SECRET_KEYbilling—Live secret key (sk_live_…) in production
STRIPE_WEBHOOK_SECRETbilling—Signing secret (whsec_…) from the same Stripe account and mode as the secret key
STRIPE_PRICE_TEAMbilling—Price ID for the $4/mo Team plan (tax-inclusive)
STRIPE_PRICE_TEAMPLUSbilling—Price ID for the $6/mo Team Plus plan (tax-inclusive). The name has no underscore between TEAM and PLUS
STRIPE_PRICE_SCALEbilling—Price ID for the $12/mo Scale plan (tax-inclusive)
STRIPE_PRICE_TEAM_ANNUALbilling—Price ID for yearly Team billing, $40/yr (tax-inclusive)
STRIPE_PRICE_TEAMPLUS_ANNUALbilling—Price ID for yearly Team Plus billing, $60/yr (tax-inclusive). No underscore between TEAM and PLUS, as in STRIPE_PRICE_TEAMPLUS
STRIPE_PRICE_SCALE_ANNUALbilling—Price ID for yearly Scale billing, $120/yr (tax-inclusive)
EMAIL_BACKENDprodconsoleSet django.core.mail.backends.smtp.EmailBackend; without SMTP no new user can verify their email, mint keys or approve CLI logins (veilfile.W001)
EMAIL_HOST, EMAIL_PORT, EMAIL_USE_TLS, EMAIL_HOST_USER, EMAIL_HOST_PASSWORDprod—SMTP settings (the live service uses Gmail SMTP)
DEFAULT_FROM_EMAILprod—From address, e.g. support@veilfile.com
EMAIL_TIMEOUTno10SMTP timeout (seconds)
VEILFILE_SAFETY_MODEnooffoff, log or enforce: SHA-256 blocklist screening of uploads. Unset is reported as veilfile.W002
VEILFILE_CLIENT_IP_HEADERnoCF-Connecting-IPHeader carrying the visitor IP, set (and overwritten) by Cloudflare. Used for rate limits when it holds a valid IP and the request came from a Cloudflare edge address; otherwise the X-Forwarded-For hop logic below applies. Empty disables it
VEILFILE_TRUSTED_PROXY_HOPSno2Trusted public proxy hops in front of the app. The client IP used for rate limits is the X-Forwarded-For entry that many positions from the right, after trailing private-network hops are skipped, when the last of them is a Cloudflare edge address. A request without one did not come through Cloudflare, and the last entry (the one the platform's proxy appended) is used. Confirm against real traffic before changing
DJANGO_SECURE_HSTS_SECONDSno31536000HSTS max-age on HTTPS responses
DJANGO_SECURE_HSTS_INCLUDE_SUBDOMAINS, DJANGO_SECURE_HSTS_PRELOADnoFalseTurn on only after every subdomain serves HTTPS
PORTno8000Set by the host (Render); the entrypoint falls back to 8000
WEB_CONCURRENCYno1gunicorn workers (gthread)
GUNICORN_THREADSno4Threads per worker. Each in-flight upload is held in memory; lower this on small instances
GUNICORN_TIMEOUT, GUNICORN_GRACEFUL_TIMEOUTno120, 30gunicorn timeouts (seconds)
SKIP_MIGRATEnounset1 skips migrate at boot (use with a Render pre-deploy command)
VEILFILE_FUNNEL_RETENTION_DAYSno365How long prune_funnel_events keeps funnel rows
VEILFILE_ALERT_DIRnounsetDirectory where CRITICAL safety alerts are also written as files, for an external alert bus; they always go to the log
DJANGO_TEST_ALLOW_REMOTE_DBno—Lets manage.py test run against a non-local database server (CI with a service container)

Without Stripe vars, checkout/switch/portal return 503 billing_not_configured and the dashboard shows a friendly note; everything else works.

Tests

Tests need DJANGO_DEBUG=True (the test runner refuses False) and a local database; it also forces local storage and strips STRIPE_*/AWS_*.

# Backend (the count changes; run it rather than trusting a number here)

cd backend && DJANGO_DEBUG=True DJANGO_SECRET_KEY=dev-only-key \ ../venv/bin/python manage.py test ../venv/bin/python manage.py makemigrations --check --dry-run # must report no changes

CLIs

cd npm && node --test cd cli && ../venv/bin/python -m unittest discover -s tests -t .

CI (.github/workflows/ci.yml) runs all of the above plus manage.py check --deploy, a production-settings collectstatic, and the browser checks in site-tests/ (jobs site, site-cloth, site-ink, site-reading, site-phone, site-forms, site-catalog and site-dashboard; see AGENTS.md, "Site checks") on every pull request to main (the browser checks skip a pull request that changes no page), and the backend and CLI checks on every push to it. It runs on GitHub-hosted runners, one machine for each job, with the jobs side by side (AGENTS.md, "Where CI runs").

Ops

  • Deploy: Render Docker web service (serves veilfile.com), auto-deploys every commit to main after its CI checks pass, with a paid Postgres. render.yaml mirrors it. The container runs as a non-root user. Boot sequence (backend/entrypoint.sh): migrate --noinput (skipped when SKIP_MIGRATE=1), createcachetable, a background daily housekeeping loop, then gunicorn.
  • Health check: GET /healthz returns 200 {"status":"ok"} after a SELECT 1, or 503. Set it as the Render Health Check Path.
  • Daily housekeeping: the entrypoint loop runs sweep_expired, clearsessions, cleartokens, prune_funnel_events, prune_email_ips (clears send-log IPs after 90 days) and purge_expired_cache once a day. No separate cron job is needed. Uploads also trigger a small lazy sweep.
  • TTL sweeper: python backend/manage.py sweep_expired deletes expired artifacts (files + rows). Idempotent. Prints swept=N bytes=N failed=N and exits non-zero when any storage delete failed (those rows are kept for the next run); --limit N / --time-budget SECONDS bound a run.
  • Orphaned files: python backend/manage.py reconcile_storage lists stored objects with no artifact row; --delete removes them. Run it by hand, dry run first: --delete refuses when the database looks wrong (empty artifact table, more than 20% orphans, or R2 storage without PostgreSQL) unless you pass --force after checking the db= / storage= it prints.
  • Billing resync: python backend/manage.py sync_stripe_billing (dry run; --apply to write) rebuilds workspace plans from Stripe. The Stripe setup steps are in the private ops repository.
  • Deploy checks: manage.py check --deploy reports veilfile.W001-W006 misconfigurations (also logged at ERROR by every worker at boot).
  • Local container: docker run needs DATABASE_URL=sqlite:////tmp/db.sqlite3 together with DJANGO_ALLOW_SQLITE=True (production mode otherwise refuses SQLite); /app/backend is read-only to the non-root user.
  • Storage: Render's disk is ephemeral. Production must use `DROP_STORAGE=s3` (Cloudflare R2).
  • MCP registries: launch checklist in docs/REGISTRIES.md.

Project layout

backend/

config/ Django settings/URLs (env-driven) drop_core/ Workspace, APIKey (vlf_), Artifact, plans, storage, secret scan, agent API, key mgmt, device login, OAuth glue, sweeper drop_billing/ Stripe checkout/switch/portal/status/webhook (live,

只显示了开头部分;其余内容请在仓库中阅读

问 PopNx