debuganything (Python)

Debug Anything from Python. Observe a running process, container or pod — syscalls, CPU profiles, thread stacks, sockets, packets, heap retainers — and get findings that cite evidence. This package wraps the DebugAnything CLI (Node ≥ 22 required: npx -y debuganything is used automatically) and ships tool adapters for LangChain, LlamaIndex and CrewAI.

pip install debuganything                 # core (no deps)
pip install 'debuganything[langchain]'    # + LangChain StructuredTools
pip install 'debuganything[llamaindex]'   # + LlamaIndex FunctionTools
pip install 'debuganything[crewai]'       # + CrewAI BaseTools

Plain Python

from debuganything import DebugAnything

debuganything = DEA()                       # uses `debuganything` if installed, else `npx -y debuganything@latest`
print(da.readiness()["findings"])
s = da.attach(12345)             # pid, "my-container", "ns/pod", or a command
before = da.snapshot_memory(s)
# ... exercise the target ...
after = da.snapshot_memory(s)
diff = da.diff_runs(s, before["run_id"], after["run_id"])
for line in DebugAnything.findings(diff):
    print(line)                   # [high] heap_retainer: name=Array grew 2.0 KB → 50.1 MB (observed; e4,e1)
da.report(s, out="report.md")
da.detach(s)

Every call returns the Debug Anything ToolResult dict (ok, findings, evidence, limitations, suggested_next, artifacts). capture_traffic needs consent=True — ask your user first.

LangChain

from debuganything.langchain import dea_tools
from langgraph.prebuilt import create_react_agent

agent = create_react_agent(llm, dea_tools())
agent.invoke({"messages": [("user", "pid 12345 is slow — attach and tell me why, with evidence")]})

LlamaIndex

from debuganything.llamaindex import dea_tools
from llama_index.core.agent import ReActAgent

agent = ReActAgent.from_tools(dea_tools(), llm=llm, verbose=True)

CrewAI

from debuganything.crewai import dea_tools
from crewai import Agent

sre = Agent(role="SRE", goal="Diagnose the hung worker with evidence, never guess", tools=dea_tools(), llm=llm)

Or just use MCP

Any framework with MCP support (LangChain langchain-mcp-adapters, LlamaIndex McpToolSpec, CrewAI MCPServerAdapter, Claude Code, Codex, Cursor…) can run Debug Anything directly: {"command": "npx", "args": ["-y", "debuganything@latest", "serve"]}. Same 14 tools, same ToolResult, plus the guided prompts.

Safety

Debug Anything is read-only, never runs sudo, caps every tool at 60 s, redacts secrets in what it returns, and keeps raw artifacts under ~/.debuganything/sessions/. See docs/safety.md.

License and disclaimer

MIT. Provided "as is", without warranty of any kind; the authors accept no liability for any damage arising from use. Only observe systems you own or are explicitly authorized to debug — you are solely responsible for authorization and legal compliance. Findings are automated diagnostics, not professional advice. See the full DISCLAIMER.