Add Valmera as an MCP Server

Streamable HTTP · OAuth 2.1 with dynamic client registration and PKCE · a bearer-token path for clients without OAuth · 143 tools (live server card, October 2026)

CONFIG LOCATION

Customize → Connectors → + Add → Add custom connector

AUTH

OAuth — nothing to paste but the URL

Name it Valmera, paste the endpoint and continue. Claude reads the 401 challenge, discovers the authorization server and opens a Valmera sign-in and consent page. If Claude asks how it should identify itself (OAuth client), choose Register automatically: Valmera uses dynamic client registration, so there is no client ID or secret to create.

https://valmera.io/mcp/server

• Custom connectors are available on Free, Pro, Max, Team and Enterprise. Free accounts are limited to one custom connector.
• On Team and Enterprise, an Owner adds it from Organization settings → Connectors, and members click Connect under Customize → Connectors.
• Once added, the connector works across claude.ai, Claude Desktop and the mobile apps. Turn it on per chat from + → Connectors.
• Sign in with the email or Google account you subscribed with. Google accounts use the normal Valmera login; no separate password is needed.

Full Claude guide →

Checked 2026-10-06 against Anthropic's custom-connector help article.

CONFIG LOCATION

chatgpt.com/plugins → + → Create custom MCP server

AUTH

OAuth — nothing to paste but the URL

Name it Valmera and paste the endpoint under Connection. Keep OAuth as the authentication, review the risk notice, select I understand and want to continue, then Create as a plugin. Sign in to Valmera when prompted. In a new chat, type @ and choose Valmera.

https://valmera.io/mcp/server

• If the create option is missing, turn on Developer mode in Settings → Security and login. Availability depends on your ChatGPT plan and workspace policy.
• ChatGPT caches the tool list. After Valmera adds tools, open the connection on chatgpt.com/plugins, select Refresh and start a new chat.
• Files attached to a ChatGPT message don't reach Valmera. Upload in Studio, or paste a link for fetch_url to import.

Full ChatGPT guide →

Checked 2026-10-06 against OpenAI's connect-and-test guide for MCP servers.

CONFIG LOCATION

~/.claude.json (local/user scope) or .mcp.json at the repo root (project scope)

AUTH

OAuth, or a bearer token in a header

One command, then /mcp inside a session to complete the browser sign-in. Add --scope user to get it in every repo, or --scope project to write a committed .mcp.json your team inherits.

claude mcp add --transport http valmera \
  https://valmera.io/mcp/server

• Confirm with claude mcp list — a working server reads ✔ Connected, ! Needs authentication means the sign-in has not completed, ✘ Failed to connect is transport or credentials.
• For a static token instead: --header "Authorization: Bearer vlm_mcp_..." (-H is the short form).
• Hand-editing .mcp.json: an entry with a url and no type is read as a stdio server and skipped. Keep "type": "http" — "streamable-http" is accepted as an alias.
• Claude Code expands ${VAR} inside url and headers, so a committed .mcp.json can reference ${VALMERA_MCP_TOKEN} and keep the secret out of git.

Full Claude Code guide, with a session end to end →

Checked 2026-08-04 against code.claude.com/docs/en/mcp and used end to end with Valmera.

CONFIG LOCATION

~/.codex/config.toml (or .codex/config.toml in a trusted project)

AUTH

OAuth via codex mcp login, or a bearer token

Add the server, then run the login command; Codex opens the Valmera sign-in in your browser. In the IDE extension, open the gear menu → MCP servers → Add server, choose Streamable HTTP, paste the URL, restart the extension and select Authenticate.

codex mcp add valmera --url https://valmera.io/mcp/server
codex mcp login valmera

OR ADD THE TABLE TO ~/.CODEX/CONFIG.TOML BY HAND

[mcp_servers.valmera]
url = "https://valmera.io/mcp/server"

• For a token instead of OAuth, add bearer_token_env_var = "VALMERA_MCP_TOKEN" to the same table and set that variable.
• Codex has a shell, so it can upload local files with upload_start and upload_finish.

Checked 2026-10-06 against OpenAI's Codex MCP documentation.

CONFIG LOCATION

~/.cursor/mcp.json (global) or .cursor/mcp.json (this project)

AUTH

OAuth — Cursor runs the flow itself

A remote server is a url entry. Cursor discovers the authorization server, registers itself and opens the Valmera sign-in — there is no client id, secret or token to fill in. You can also add and toggle servers from Customize in the sidebar.

{
  "mcpServers": {
    "valmera": {
      "url": "https://valmera.io/mcp/server"
    }
  }
}

• One-click install: the Add to Cursor button on the connect page and the Cursor guide writes this entry for you.
• Cursor's documented remote shape is url plus optional headers; it does not require a type field.
• Cursor asks before running each MCP tool by default. Most of an edit is reading — look_at, get_words, project_state — so use a Run Mode with an allowlist or the session becomes a confirmation dialog.
• Cursor's OAuth redirect URLs are fixed: https://www.cursor.com/agents/mcp/oauth/callback for web and agents, http://localhost:8787/callback on desktop. Both are already accepted by Valmera's dynamic registration.
• MCP Logs in the Output panel (Cmd+Shift+U) separate an unreachable URL from a rejected sign-in.

Full Cursor guide, including the screen-recording tools →

Verified 2026-08-04 against cursor.com/docs/context/mcp.

CONFIG LOCATION

.vscode/mcp.json in the workspace, or the user profile via MCP: Open User Configuration

AUTH

OAuth — VS Code opens a browser on first connect

VS Code's key is servers, not mcpServers — a config copied from another client's docs will not load. Run MCP: Add Server from the command palette to have it written for you, or paste this. MCP works in agent mode.

{
  "servers": {
    "valmera": {
      "type": "http",
      "url": "https://valmera.io/mcp/server"
    }
  }
}

• MCP: List Servers shows status; MCP: Browse MCP Servers opens the gallery.
• For a static token, add a headers object. Do not hardcode it — use the inputs array with ${input:valmera-token} and type promptString with password: true, so VS Code prompts and stores it instead of committing it.
• An explicit oauth block with a clientId exists for servers that require a pre-registered client. Valmera does not — omit it and let VS Code register dynamically.

Checked 2026-08-04 against code.visualstudio.com's MCP server and MCP configuration reference docs.

CONFIG LOCATION

The MCP Servers icon → Remote Servers tab, or ~/.cline/mcp.json for the CLI

AUTH

Bearer token — no documented OAuth flow

Add it from the Remote Servers tab: a name, the URL, and Streamable HTTP as the transport. Cline's documented remote authentication is a header, so mint a Valmera bearer token first — see below. The type field is streamableHttp, camel-cased, which is Cline's own spelling.

{
  "mcpServers": {
    "valmera": {
      "type": "streamableHttp",
      "url": "https://valmera.io/mcp/server",
      "headers": {
        "Authorization": "Bearer vlm_mcp_..."
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

• Cline's docs describe header authentication and do not document an OAuth browser flow for remote servers. If you would rather not hold a token, wrap the endpoint in mcp-remote and add it as a stdio server instead.
• autoApprove is a per-tool allowlist. Leave it empty until you know which calls you want unattended: renders and pixel repair are real compute.

Checked 2026-08-04 against docs.cline.bot.

CONFIG LOCATION

~/.codeium/windsurf/mcp_config.json

AUTH

OAuth is supported; a header also works

A remote server takes serverUrl (url is also accepted). Headers support ${env:VAR} and ${file:/path} interpolation, so the token stays out of the file.

{
  "mcpServers": {
    "valmera": {
      "serverUrl": "https://valmera.io/mcp/server",
      "headers": {
        "Authorization": "Bearer ${env:VALMERA_MCP_TOKEN}"
      }
    }
  }
}

• Read this before you connect: Cascade uses at most 100 tools at a time across all servers, and Valmera exposes 143. List the tools you don't need in the server's disabledTools array in mcp_config.json. No other client on this page has that ceiling.
• Windsurf supports OAuth for each transport type, so the headers block above is optional if the sign-in flow works for you.

Checked 2026-10-06 against Windsurf's Cascade MCP docs.

CONFIG LOCATION

~/.config/zed/settings.json (macOS/Linux), %APPDATA%\Zed\settings.json (Windows)

AUTH

OAuth when no Authorization header is set

Zed calls MCP servers context servers. Settings → AI → MCP Servers → Add Server → Add Remote Server writes this for you. When a remote server has no configured Authorization header, Zed runs the standard MCP OAuth flow.

{
  "context_servers": {
    "valmera": {
      "url": "https://valmera.io/mcp/server"
    }
  }
}

• Native remote-server support is recent. Older Zed builds had no url form at all and the workaround was mcp-remote as a stdio context server — if your settings file rejects url, that is the version you are on, and the fallback at the bottom of this page is the answer.
• A headers object with Authorization: Bearer works if you would rather use a token.

Checked 2026-08-04 against zed.dev/docs/ai/mcp.

CONFIG LOCATION

.continue/config.yaml in the workspace, or a file under .continue/mcpServers/

AUTH

Bearer token — OAuth is not documented

The type is streamable-http, hyphenated and case-sensitive — streamableHttp and http are both rejected. Authentication goes in requestOptions.headers, which Continue documents for sse and streamable-http servers.

mcpServers:
  - name: valmera
    type: streamable-http
    url: https://valmera.io/mcp/server
    requestOptions:
      headers:
        Authorization: Bearer vlm_mcp_...

• MCP is available in agent mode only.
• Continue's config reference does not mention OAuth for MCP servers, so plan on a token.

Checked 2026-08-04 against docs.continue.dev.

CONFIG LOCATION

~/.config/goose/config.yaml (macOS/Linux), %APPDATA%\Block\goose\config\config.yaml (Windows)

AUTH

OAuth on a 401 challenge, or a header

Run goose configure → Add Extension → Remote Extension (Streaming HTTP), name it, and paste the endpoint as the URI. Use the interactive path for the first setup — it writes correct YAML — and hand-edit afterwards to change the timeout or add headers.

extensions:
  valmera:
    enabled: true
    type: streamable_http
    name: valmera
    description: Agentic video editor
    uri: https://valmera.io/mcp/server
    timeout: 60

• Two spellings are Goose's alone: the key is uri where every other client uses url, and the type is streamable_http with an underscore.
• Goose triggers OAuth when a server answers 401 with a WWW-Authenticate header, which is exactly what Valmera answers. Dynamic client registration is what makes that work without a client id.
• Raise timeout above the default if you script long sequences; Valmera's own calls return within about 25 seconds regardless.

Checked 2026-08-04 against Goose's extension configuration docs.

CONFIG LOCATION

librechat.yaml, under a top-level mcpServers block

AUTH

OAuth with dynamic client registration, or headers

The simple form is a type, a url and a timeout in milliseconds. LibreChat is a multi-user server, so the interesting question is whose Valmera account gets used — see the second block.

mcpServers:
  valmera:
    type: streamable-http
    url: https://valmera.io/mcp/server
    timeout: 60000

PER-USER OAUTH INSTEAD OF A SHARED TOKEN

mcpServers:
  valmera:
    type: streamable-http
    url: https://valmera.io/mcp/server
    oauth:
      authorization_url: https://valmera.io/mcp/oauth/authorize
      token_url: https://valmera.io/mcp/oauth/token
      redirect_uri: http://localhost:3080/api/mcp/valmera/oauth/callback

• Per-user identity is the right shape here. An oauth block with client_id and client_secret omitted makes LibreChat use dynamic client registration, so each of your users signs into their own Valmera account rather than sharing one.
• The alternative — one bearer token in a static header — means every LibreChat user edits the same Valmera projects on one account. That is occasionally what you want and usually not.
• customUserVars prompts each user for their own credential and interpolates it as {{VAR}} in headers, which is the middle path if you are issuing tokens yourself.

Checked 2026-08-04 against librechat.ai's mcp_servers reference. Endpoint paths are Valmera's own, from its authorization-server metadata.

CONFIG LOCATION

Nothing to install — npx, Node 22.19.0 or newer

AUTH

Full OAuth, or --header

This is the one to reach for when a client will not connect and you need to know whose fault it is. The Inspector implements the whole authorization flow, so if it connects and lists Valmera's tools, the server is fine and the problem is the client's config.

npx @modelcontextprotocol/inspector \
  --server-url https://valmera.io/mcp/server --transport http

• Add --cli for a scriptable client and --tui for a terminal UI; all three share the same OAuth state on disk.
• List the tools without a browser step, once a token is stored: npx @modelcontextprotocol/inspector --cli --server-url <endpoint> --transport http --method tools/list
• A static token instead of the browser flow: --header "Authorization: Bearer vlm_mcp_...", repeatable.
• The CLI and TUI complete OAuth on http://127.0.0.1:6276/oauth/callback; the web client uses http://localhost:6274/oauth/callback. Both are accepted by Valmera's dynamic registration.
• In CI, pass --stored-auth-only so it fails immediately instead of waiting for a callback nobody will complete.

Checked 2026-08-04 against the MCP Inspector docs.

CONFIG LOCATION

.gemini/settings.json in the project (the default), or ~/.gemini/settings.json with -s user

AUTH

OAuth via /mcp auth

Add the server with one command, then authenticate from inside Gemini CLI. These steps are for Gemini CLI, not the Gemini chat website.

gemini mcp add --transport http valmera https://valmera.io/mcp/server
# then, inside Gemini CLI:
/mcp auth valmera

• For a token instead of OAuth, add --header "Authorization: Bearer vlm_mcp_..." to the add command.
• Gemini CLI has a shell, so it can upload local files with upload_start and upload_finish.

Checked 2026-10-06 against the Gemini CLI MCP server docs.

Made with Valmera: real exports