# execkit

execkit gives AI agents persistent shell sessions across local machines, SSH hosts, and Docker. Use it when structured results, command records, or remote file restore matter.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 12
- Author: jayashankarvr
- License: Apache-2.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/jayashankarvr/execkit
- Updated: 2026-09-27
- HTML: https://affpapa.org/en/nevibecode/mcp/execkit

## What it does

execkit keeps shell state between calls, including the working directory and environment. It can run sessions locally, over SSH, or inside Docker. Each execution returns separate stdout and stderr, an exit code, duration, and cwd, so the agent can work with parsed results instead of terminal text.

The server filters common secret formats before output reaches the model and limits noisy results with tailing, head output, grep, or a character cap. A policy fence checks commands before execution. Sessions can write JSONL audit records, appear in `execkit-mcp watch`, and support checkpoints for restoring files in remote workspaces.

Setup uses `uvx`. Commands time out after 120 seconds by default, with `timeout_secs` available up to 3600. The API is still in `0.x`, so a minor-version update can require a configuration check.

## Requirements
- Needs: Python and uv (commands run via uvx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add execkit -- uvx execkit-mcp
```
uvx runs the command named after the package; if the README names another one, use uvx --from execkit-mcp <command>

### Codex

**Platform CLI**
```bash
codex mcp add execkit -- uvx execkit-mcp
```
uvx runs the command named after the package; if the README names another one, use uvx --from execkit-mcp <command>

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "execkit": {
            "command": "uvx",
            "args": [
                "execkit-mcp"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from execkit-mcp <command>

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "execkit": {
            "type": "stdio",
            "command": "uvx",
            "args": [
                "execkit-mcp"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from execkit-mcp <command>

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add execkit uvx execkit-mcp
```
uvx runs the command named after the package; if the README names another one, use uvx --from execkit-mcp <command>

## When to use it
- keep `cwd` and environment across shell calls
- return stdout, stderr, and exit status separately
- redact secrets before the agent sees output
- record agent commands in a JSONL audit log
- restore files in a remote workspace

## Example prompts
- Create an execkit local session, move into `app`, and run the build commands.
- Use execkit to connect over SSH and return the command result with its exit code.
- Run the build inside Docker, cap the output, and keep the session available for audit.

## Similar
- [create-architecture-documentation](https://affpapa.org/en/nevibecode/commands/create-architecture-documentation.md) — command · 6136 installs
- [project-references](https://affpapa.org/en/nevibecode/skills/project-references.md) — skill · 3900 installs
- [api-review](https://affpapa.org/en/nevibecode/skills/api-review.md) — skill · 3800 installs
- [camel-matrix](https://affpapa.org/en/nevibecode/skills/camel-matrix.md) — skill · 3500 installs
- [web-performance-optimization](https://affpapa.org/en/nevibecode/skills/web-performance-optimization.md) — skill · 3490 installs
- [tigris-s3-migration](https://affpapa.org/en/nevibecode/skills/tigris-s3-migration.md) — skill · 3100 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md