Argus — API and MCP reference

Documentation, the assistant skill briefs, and the MCP registry manifest for Argus, an anti-detect browser for managing hundreds of isolated browser profiles — each with its own fingerprint, proxy and cookie jar.

Which Argus this is. Argus is published by Simnetiq Ltd (UK company number 16861177) at www.browserargus.com. Other companies and software products share the name Argus and are unrelated to this one.

This is a local API. It listens on loopback on the machine running Argus, and nothing off that machine can reach it. There is no hosted web API.

There is nothing here to install

The Argus MCP server ships inside the launcher's own application bundle and is started through it with ELECTRON_RUN_AS_NODE. There is no npm package, no npx command and nothing new listening: it speaks JSON-RPC over stdin and stdout to the client that launched it, and reaches the local API on your behalf.

For most clients the launcher writes the config file for you and shows you the path first. Two of them have no file it can write, and hand you a snippet instead. See Integrations and MCP setup.

This repository is the documentation for that server, not the server.

What is here

Quick start

1. Download Argus and open the launcher.
2. Mint a key in the API tab. It is shown once. Export it as ARGYS_API_TOKEN — that is the name the launcher tells you to use and the one the MCP server reads from the environment.
3. Call it:

curl -s "http://127.0.0.1:39219/v1/profiles" \
  -H "Authorization: Bearer $ARGYS_API_TOKEN"

The first call from a new client raises an approve-or-deny card in the launcher and blocks until you answer it.

To drive it from a coding agent instead, connect an MCP client — Claude Code, Codex, Cursor, Gemini CLI, Windsurf, VS Code, Zed, OpenClaw, Hive and others are supported — and the 143 tools appear on their own.

About skills/

An Argus skill is the working brief the in-app assistant loads for one job: instructions, worked examples, a reference shelf, and the pack of tools it may reach.

Every brief in skills/ already ships inside Argus. They are published here to be read and to be adapted — not installed, because you already have them. Each file is in the shape POST /v1/skills/save accepts, so editing one and posting it creates your own custom skill:

curl -X POST "http://127.0.0.1:39219/v1/skills/save" \
  -H "Authorization: Bearer $ARGYS_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d @skills/proxies.json

skills/README.md is the roster: every brief, what it is for, and which pack of tools each one carries.

Note what the files deliberately omit:

• No skillId. Posting one without an id creates a new custom skill. Posting a built-in's id would instead write a local override of that built-in, which is not what a reader of this repo is asking for.
• No entryTabs. A custom skill can claim a tab from a built-in, and an installed copy silently displacing the shipped skill's tab suggestion is a surprise nobody asked for.
• toolPack only where one fits. A skill gets its tools by naming one of the packs Argus defines — naming individual tools is not possible by design, so that a skill document can never grant itself a capability the code does not already ship. Several briefs use a tool set that is not any one pack; their files carry no toolPack, so posting one as-is gives a skill with documentation and navigation tools only. The brief is still the useful part. skills/README.md marks which is which, brief by brief.

Links

• Website: https://www.browserargus.com
• API reference: https://www.browserargus.com/api-reference
• Integrations and MCP setup: https://www.browserargus.com/integrations
• Everything as one file: https://www.browserargus.com/llms-full.txt

Licence

MIT — see LICENSE. The grant covers the contents of this repository: the generated documentation, the agent rules and the skill briefs. Argus itself — the launcher, the hardened browser and the MCP server — is proprietary and is not part of this repository.