Verlon AI MCP

ai.verlonPublished by verlon.aiv0.4.1Updated Oct 24, 2026

Inspect and manage Verlon AI gates, logs, recommendations, and experiments from any MCP client.

VerifiedSTDIODesktop onlyDeveloper ToolsAI & ML

Overview

AI-generated overview

Verlon AI MCP lets an assistant inspect your Verlon AI gates, request logs, recommendations, experiments and models, and optionally switch a gate's model.

What you can do
List and read gates, including full configuration, fallback chain, spending limits and sub-gates, by gate ID or across the account. Query recent request logs filtered by gate, time, success and limit, with cost and latency. Fetch Cortex intelligence reports with themes, drift detection and optimization recommendations. List shadow and split experiments with status, variants and goal metrics. List available chat models by provider with live pricing and capability scores. With writes enabled, switch the model a gate routes to.
When to use it
Useful when you work with Verlon AI and want a coding agent to look up gate configuration, recent request costs and failures, or optimization recommendations without leaving the editor. Add it if you want the assistant to change a gate's routing model on request.
Requirements
Runs locally over stdio, spawned by your MCP client via npx, so Node.js and network access to the Verlon API are needed. A Verlon API key is required in the VERLON_API_KEY environment variable. VERLON_BASE_URL optionally overrides the API base URL for self-hosted Verlon. Write tools require starting the server with the --enable-writes flag.
Before you install
The server holds a Verlon API key (VERLON_API_KEY) that grants access to your account; keep it out of shared configs. Read-only by default, but with --enable-writes the switch_model tool changes which model a gate routes to, taking effect on the next turn of running sessions. The README notes broader write tools are planned behind the same flag. The assistant may call tools autonomously when a request seems relevant.

Installation

In SourceWeft

  1. Open Verlon AI MCP in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

@verlon-ai/mcp

[npm version] [license] [CI]

Model Context Protocol server for Verlon AI. Exposes your Verlon resources (gates, logs, recommendations, experiments) as MCP tools so coding agents — Claude Code, Cursor, Cline, any MCP-compatible client — can inspect and manage your AI infrastructure natively.

Status: 0.4.0 — listed in the MCP Registry as ai.verlon/mcp. Ships 6 read-only tools (list_gates, get_gate, list_logs, get_recommendations, list_experiments, list_models) plus one write tool, switch_model, registered only with --enable-writes. The broader write surface (create_gate, update_gate, run_chat, start_experiment) lands in a future release behind the same flag.

Install

You don't install it directly. Your MCP client (Claude Code, Cursor, etc.) spawns it as a subprocess via npx. Add the snippet below to your client's MCP config.

Claude Code

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on your OS:

json
{  "mcpServers": {    "verlon": {      "command": "npx",      "args": ["-y", "@verlon-ai/mcp"],      "env": {        "VERLON_API_KEY": "sk-vrln-..."      }    }  }}

Then restart Claude Code. The verlon server should appear in the tools list, and Claude can call verlon:list_gates against your account.

Cursor

Add to your Cursor MCP config (Settings → Features → MCP Servers):

json
{  "mcpServers": {    "verlon": {      "command": "npx",      "args": ["-y", "@verlon-ai/mcp"],      "env": {        "VERLON_API_KEY": "sk-vrln-..."      }    }  }}

Any other MCP-compatible client

The server speaks MCP over stdio. Spawn npx -y @verlon-ai/mcp with VERLON_API_KEY in the subprocess environment.

Tools

The default tool set is read-only — see Security note for the rationale. Write tools register only when the server starts with --enable-writes.

ToolInputsWhat it returns
list_gatesnoneEvery gate in the account — id, name, description, model, taskType, taskSubtype, createdAt
get_gategateId (UUID)Full gate config — model, fallback chain, task type, spending limits, sub-gates, orchestration
list_logsgate?, since? (ISO 8601), success?, limit? (1-100, default 20)Recent request logs — timestamp, gate, model, cost, latency, success/failure
get_recommendationsgateId (UUID)Cortex intelligence report — themes, drift detection, optimization recommendations. { report: null } when no run has been produced yet
list_experimentsgateId?, status?, projectId?Experiments (shadow + split) — id, name, status, test type, variants, goal metric, configuration
list_modelsprovider? (openai, anthropic, google, mistral, …)Chat models a gate can route to, with live pricing (USD per 1M tokens) and capability scores

Write tools (--enable-writes only)

ToolInputsWhat it does
switch_modelmodel (id from list_models), gateId?Switches which model a coding gate routes to. Takes effect on the next turn of any running session, no restart. With gateId omitted, targets the account's Claude Code connector gate. Idempotent; the change is one reversible field.

Configuration

Env varRequiredDefaultNotes
VERLON_API_KEYYes—Your Verlon API key (sk-vrln-...).
VERLON_BASE_URLNohttps://api.verlon.aiOverride for self-hosted Verlon.

CLI flags

FlagPurpose
--enable-writesRegister write-capable tools (switch_model today). Default is read-only — a misaligned agent can't accidentally destroy resources.
--help, -hPrint usage.

Security note

Read-only by default is a deliberate choice. The MCP client (Claude Code, Cursor, etc.) sees this server's tools and may invoke them autonomously when a user's request makes them seem relevant. A read-only default means even a misaligned agent can only inspect your account, not modify it. Opt in to write tools (--enable-writes) only after you understand the implications. The only write tool today is switch_model, deliberately the narrowest possible first write: one reversible field on one gate. Creating, updating, or deleting resources is not yet exposed.

Development

bash
npm installnpm test          # vitestnpm run build     # tsc → dist/

Publishing (maintainers)

The package is dual-published: to npm as @verlon-ai/mcp (automated, with provenance), and to the MCP Registry as ai.verlon/mcp (manual). The registry validates that the npm version exists before accepting a publish, so npm always goes first.

Per-release flow

Bump versions in lockstep across three files — CI fails on drift:

FileField
package.jsonversion
server.jsonversion AND packages[0].version
src/server.tsVERLON_MCP_VERSION constant

Then:

bash
# 1. Merge the bump to main (CI enforces the lockstep), then tag:git tag v0.4.1 && git push origin v0.4.1# The publish workflow runs `npm publish --provenance` automatically.
# Wait ~30s for npm CDN; verify:npm view @verlon-ai/mcp version   # should print the new version
# 2. MCP Registry publish (manual — needs mcp-publisher + DNS-verified ai.verlon namespace)npm run publish:mcp

One-time setup (registry publishing)

bash
# Install the MCP Registry publisher (NOT npm — it's a prebuilt binary)brew install mcp-publisher
# DNS-verify the verlon.ai domain (required to publish under the ai.verlon namespace)mcp-publisher login --help   # follow the DNS verification flow it prints# Add the TXT record on verlon.ai; verify with `dig TXT verlon.ai +short`

Verification

bash
# All three versions match?npm view @verlon-ai/mcp versionjq -r .packages[0].version server.jsongrep VERLON_MCP_VERSION src/server.ts
# Registry listing live?curl 'https://registry.modelcontextprotocol.io/v0/servers?search=verlon' | jq
# End-to-end smoke against the published artifactVERLON_API_KEY=sk-vrln-... npx @modelcontextprotocol/inspector npx -y @verlon-ai/mcp

License

MIT — see LICENSE.

Source: README.md at commit 15aaf0a

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.4.1LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts