bash-command-mcp
Enables executing bash commands with foreground/background support, timeouts, environment control, and OpenTelemetry observability.
README
bash-command-mcp
A highly sophisticated Bash MCP server for safe, structured command execution with first-class background job orchestration.
Important Security Warning
This server executes shell commands on the machine where it is running.
If you run bun run index.ts directly on your host, commands run on your host with your user permissions.
Use Docker to isolate execution unless you fully trust the MCP client and prompts.
To install dependencies:
bun install
To run over stdio:
bun run index.ts
To run over Streamable HTTP:
BASH_COMMAND_MCP_TRANSPORT=http \
BASH_COMMAND_MCP_HOST=127.0.0.1 \
BASH_COMMAND_MCP_PORT=3000 \
bun run index.ts
To run via npm/npx (published package):
npx -y bash-command-mcp
Why This Server
- High-fidelity shell execution with clear exit-code semantics.
- Advanced background process lifecycle controls (
run_background,wait_background,kill_background). - Built-in observability via per-process stdout/stderr log files.
- OpenTelemetry traces and metrics for production visibility.
- Agent-friendly ergonomics with
cwdandenvoverrides for precise execution context.
Tool Behavior
Tools:
run: run command in foreground. Args:commandorcmd,timeoutSeconds(default60, min1; values above86400are capped with a hint), optionalcwd, optionalenv.run_background: start command in background with stdout/stderr written to log files. Args:commandorcmd, optionalcwd, optionalenv.list_background: list tracked background processes, including log file paths.kill_background: stop tracked background process bypid.tail_background: show last N lines from background process logs. Args:pid, optionallines(default200, max5000).wait_background: wait for background process completion and return final status/output. Args:pid, optionaltimeoutSeconds(default60, min1; values above86400are capped with a hint).
OpenTelemetry
This server includes built-in OpenTelemetry instrumentation for traces and metrics.
- OpenTelemetry packages are installed with the server package.
- Telemetry initializes unless
OTEL_ENABLED=false. - If
OTEL_EXPORTER_OTLP_ENDPOINTis set, traces/metrics are exported via OTLP HTTP. - If no OTLP endpoint is configured, console exporters are used.
Instrumented operations:
- Tool call spans for
run,run_background,list_background,tail_background,wait_background, andkill_background. - Background lifecycle spans/counters (
started,ended). - Metrics for tool calls, failures, timeouts, and duration histograms.
Common env vars:
OTEL_ENABLED=true|falseOTEL_SERVICE_NAME=bash-command-mcpOTEL_SERVICE_VERSION=1.0.0OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318OTEL_METRIC_EXPORT_INTERVAL_MS=10000BASH_COMMAND_MCP_LOG_DIR=/path/to/log-dir
Example (OTLP Collector on localhost):
OTEL_ENABLED=true \
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 \
OTEL_SERVICE_NAME=bash-command-mcp \
npx -y bash-command-mcp
Transports
This server supports two transports:
stdiofor local, process-spawned integrations.httpfor remote or network-accessible MCP clients using Streamable HTTP.
Default behavior remains stdio so existing desktop and CLI setups keep working.
Environment variables for HTTP mode:
BASH_COMMAND_MCP_TRANSPORT=http|stdioBASH_COMMAND_MCP_HOST=127.0.0.1BASH_COMMAND_MCP_PORT=3000BASH_COMMAND_MCP_ALLOWED_HOSTS=localhost,127.0.0.1,[::1]
HTTP mode uses host-header validation by default when bound to a loopback address. If you bind to 0.0.0.0 or ::, provide an explicit allow-list in BASH_COMMAND_MCP_ALLOWED_HOSTS.
Docker
Build the image:
docker build -t bash-command-mcp .
Run with a local folder mounted at /workspace:
docker run --rm -i -v "$(pwd):/workspace" bash-command-mcp
Run over Streamable HTTP:
docker run --rm -p 3000:3000 \
-e BASH_COMMAND_MCP_TRANSPORT=http \
-e BASH_COMMAND_MCP_HOST=0.0.0.0 \
-e BASH_COMMAND_MCP_PORT=3000 \
bash-command-mcp
/workspace mapping explained:
- Left side (
$(pwd)) is a folder on your host machine. - Right side (
/workspace) is the path inside the container. - Commands run by this MCP server should target files under
/workspace; those changes are written back to the mapped host folder.
For HTTP mode in Docker, bind to 0.0.0.0 and publish the port with -p. If you expose the container beyond localhost, set BASH_COMMAND_MCP_ALLOWED_HOSTS to the hostnames you want to permit.
Example:
- If your host has
./project/file.txtand you run the container from./project, the same file is available in the container at/workspace/file.txt.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.