ModelsAtlas
Explore
Atlas / MCP servers / aplavin / Julia

JuliaSAFE

mcp/aplavin/julia
MCP server by aplavin · ai · julia · llm · Audited 2026-10-07
⚑ Report / re-scanSource ↗

MCP server for persistent Julia sessions — fast iteration without startup/compilation overhead

Verdict
SAFE
Grade
B
Trust score
89 /100
Exposed tools
3 1r · 2w · 0d
Transport
stdio
License
MIT
Stars
90
01

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

MCP server that gives AI assistants access to efficient Julia code execution. Avoids Julia's startup and compilation costs by keeping sessions alive across calls, and persists state (variables, functions, loaded packages) between them — so each iteration is fast.

  • Sessions start on demand, persist state between calls, and recover from crashes — no manual management
  • Each project directory gets its own isolated Julia process
  • Pure stdio transport — no open ports or sockets

Tools

  • julia_eval(code, env_path?, timeout?) — execute Julia code in a persistent session. env_path sets the Julia project directory (omit for a temporary session). timeout defaults to 60s and is auto-disabled for Pkg operations.
  • julia_restart(env_path?) — restart a session, clearing all state. If env_path is omitted, restarts the temporary session.
  • julia_list_sessions — list active sessions and their status

Requirements

  • uv (you might already have it installed)
  • Julia – any version, julia binary must be in PATH
  • Recommended packages – used automatically if available in the global environment:
  • Revise.jl - to pick code changes up without restarting
  • TestEnv.jl — to properly activate test environment when env_path points to /test/

The server itself is written in Python since the Python MCP protocol implementation is very mature.

Usage

First, clone the repository:

cd /any_directory
git clone https://github.com/aplavin/julia-mcp.git

Then register the server with your client of choice (see below).

That's it! Your AI assistant can now execute Julia code more efficiently, saving of TTFX.

Claude Code

User-wide (recommended — makes Julia available in all projects):

claude mcp add --scope user julia -- uv run --directory /any_directory/julia-mcp python server.py

Project-scoped (only

Read from source at commit d264236d59ccOBSERVED · 2026-10-07
02

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.

claude-code
claude mcp add julia-mcp -- uvx julia-mcp
claude-desktop
{
  "mcpServers": {
    "julia-mcp": {
      "command": "uvx",
      "args": [
        "julia-mcp"
      ]
    }
  }
}
03

Exposed tools (3)

1 read · 2 write · 0 destructive.

ToolRiskDescription
julia_evalwriteALWAYS use this tool to run Julia code. NEVER run julia via command line.
julia_list_sessionsreadList all active Julia sessions and their environments.
julia_restartwriteRestart a Julia session, clearing all state.
04

Trust audit

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

LayerWhat it checksResult
L0Provenance & inventoryPASS
L1Static analysis of the codePASS
L2Instruction surface (what it tells the agent)PASS
L3Class-specific surfacePASS
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
declared (2 observation(s))
Network
none-observed
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.

Audited 2026-10-07 · audit v0.4.1 · source sha d264236d59ccfull audit observations/trust-audit/mcp-server/aplavin__julia.json · Report an issue / request a re-scan
05

Audit history

Every audit this server has had. A grade with a past is a grade somebody is still checking.

DateSourceVerdictGradeScoreChange
2026-10-07d264236d59ccSAFEB89first audit
06

Questions

What is the Julia MCP server?

MCP server for persistent Julia sessions — fast iteration without startup/compilation overhead

What tools does Julia expose?

3 in total: 1 read-only, 2 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.

Is Julia safe to connect to an agent?

The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.

What credentials does Julia need?

No credential environment variables were found in its source, so it appears to need none.

How does Julia run?

It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as julia-mcp.

How current is this page?

The grade is for one exact copy of the source (d264236d59cc), read on 2026-10-07. The repository is watched and re-audited when it changes.

Advertisement