mcpField
A
Arnesa

Arnesa

Local
github.com/arnesa-aiOther0Updated today
Repository

About Arnesa

Arnesa is the CLI for an independent benchmark service for coding-agent harnesses. Developers can inspect a project profile, get recommendations for verified harnesses, install or remove configuration, and scan agent configuration files.

Key features

  1. 01Detect project profile
  2. 02Recommend verified harnesses
  3. 03Install or remove configuration
  4. 04Scan agent configuration files
  5. 05Output results as JSON

Use cases

  1. 01Identify harnesses suited to a project
  2. 02Get setup recommendations from project traits
  3. 03Preview and apply harness configuration changes
  4. 04Review agent configuration files

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "arnesa-ai-arnesa": {
      "command": "npx",
      "args": [
        "-y",
        "arnesa"
      ]
    }
  }
}
Run commandnpx -y arnesa

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npmarnesa0.2.0

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

arnesa

The command line client for Arnesa, the independent benchmark for AI coding-agent harnesses. Point it at your Arnesa deployment with --api <url> or the ARNESA_API_URL environment variable.

npx arnesa detect      # print the profile of this project (nothing leaves your machine)
npx arnesa recommend   # ask Arnesa which verified harnesses fit this profile
npx arnesa install <slug>
npx arnesa remove <slug>
npx arnesa doctor      # scan your agent configuration files
npx arnesa login       # sign in with a device code

Requires Node.js 22 or newer.

What leaves your machine

  • detect sends nothing.
  • recommend and install send a closed-vocabulary profile (languages, frameworks, runtimes), never file names or contents.
  • doctor lists the agent configuration files it would upload, asks first, and uploads only those. It never uploads source code, .env files or credential files. Uploaded files are deleted after the scan, within one hour at most.
  • Your token is kept in the operating system keychain, never in a file.

Safety of install

The server response is treated as untrusted. Files are written only inside the allowed agent directories, after a preview and a confirmation. Every write is backed up under .arnesa/backups/, and arnesa remove restores the project. Use --yes to confirm files and --allow <gate-id> for a command that runs on its own.

Exit codes

CodeMeaning
0Success
1Failure
2Usage error
3Confirmation required
4Refused by policy
5Rate limited

Add --json for one JSON document on stdout.

Limits

Working defaults, effective 2026-11-06, confirmed by the founder on 2026-11-07 (Open Question 1, design decision 6 of add-arnesa-local-cli). They can change, and the server answers with exit code 5 when one is reached.

Callerrecommendcompile (install)
Anonymous10 per 10 minutes5 per 10 minutes
API key (ak_) or device (ad_)200 per day50 per day

Upload caps for doctor: 200 files, 256 KiB per file, 1 MiB in all.

Licence and trademark

The code is released under the MIT licence (see LICENSE). The Arnesa name and logo are not part of that licence: you may not use them to suggest that a fork or a derived product is Arnesa or is endorsed by Arnesa.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.