
Debuganything
Lets coding agents inspect a running process and cite numbered evidence for findings
- Pricing
- Open source
- Platforms
- CLI, macOS, Linux
- Runs
- Locally
- Version
- 1.1.0
- Listing updated
- 3 days ago
About Debuganything
Debuganything is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
Debug Anything gives an agent read-only eyes on live software without breakpoints, restarts or instrumentation. It attaches to a process, container or pod, picks local tools such as strace, py-spy, perf, lsof, tcpdump, gdb or lldb and Node's inspector, runs them within hard limits, and turns their output into structured evidence. Each reading is logged with its time, the tool and the raw output on your disk, and every finding cites evidence ids and is marked observed or inferred.
The site walks through a reproducible Node.js memory leak: RSS growing from 47.2 MB to 323.0 MB, with one array holding most of the heap. Areas covered include memory, CPU, thread stacks and lock waits, sockets and CLOSE_WAIT build-ups, file descriptors, syscalls, network resets and container OOM kills. A single why command takes a pid and a symptom such as slow, leak, hang or crash and writes a ranked report.
Safety limits are explicit. It never kills, patches or sets breakpoints, never runs sudo for you and instead prints the command, asks before packet capture or opening a Node inspector, redacts common token formats from what the agent sees, and has no cloud, telemetry or update checks. It runs as an MCP server, a CLI and a Python package, and is MIT licensed.
Install
Run in a terminal
claude mcp add debuganything -- npx -y debuganythingAdd to claude_desktop_config.json
{
"mcpServers": {
"debuganything": {
"command": "npx",
"args": [
"-y",
"debuganything"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"debuganything": {
"command": "npx",
"args": [
"-y",
"debuganything"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"debuganything": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"debuganything"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"debuganything": {
"command": "npx",
"args": [
"-y",
"debuganything"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"debuganything": {
"command": "npx",
"args": [
"-y",
"debuganything"
]
}
}
}What it does
- Zero-instrumentation attach
- Observes any running process, container or pod without restarts or breakpoints.
- Numbered evidence
- Logs every reading with time, tool and raw output, and findings cite evidence ids.
- Symptom-driven diagnosis
- A why command chooses tools for slow, leak, hang, crash or triage cases.
- Read-only safety
- Never kills or patches, never runs sudo, and asks before packet capture.
- Secret redaction
- Scrubs common token, key and password formats from what the agent sees.
- Reports
- Writes a report in which every finding lists its evidence.
- Python and CLI access
- Usable from LangChain, LlamaIndex or CrewAI agents and from the command line.
Good for
- Find why a Node.js or Python server's memory keeps growing
- See where CPU time goes in a slow process
- Diagnose deadlocks and lock waits through thread stacks
- Find who holds a port or why connections pile up in CLOSE_WAIT
- Check container OOM kills and limits against usage
Pricing
Open source
Open source, MIT licensed
Prices change. Check debuganything.dev before you buy.
Debuganything FAQ
Does it need root?
Most tools do not. When one does, it prints the exact command for you to run, then reads the result with debuganything import.
Does anything leave my machine?
No. The site says there is no cloud, telemetry or update check.
Can it change or kill processes?
No. It is read-only. The one exception is on Node, where it can send SIGUSR1 to open the inspector, only after you say yes.
Which languages are supported?
Node, Python, Go, Java and native processes, plus containers and pods.
How do I add it to an agent?
Run npx debuganything setup, or add an npx command with arguments -y, debuganything, serve to your MCP config.
Is Debuganything free?
Yes. Debuganything is open source, so you can use and self-host it for free.
How do I add Debuganything to Claude Code?
Run this command in your terminal: claude mcp add debuganything -- npx -y debuganything
What are the best alternatives to Debuganything?
The closest alternatives to Debuganything are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to Debuganything

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Ouroboros
MCP serverSpec-first layer that pins requirements and verifies results of AI coding agent work
Runs locally- Open source
- 6.4K GitHub stars
- Developer tools