Skip to main content
Glama

debugpy_attach

Attach debugpy to running Python processes inside Docker containers for debugging and inspection. Enables process injection and breakpoint planning based on logs.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
containerYes
pidNo
portNo
hostNo0.0.0.0
python_binNopython
wait_for_clientNo
log_toNo/tmp/debugpy-logs
configure_subprocessNo

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

Implementation Reference

  • The main handler function for the 'debugpy_attach' tool. It checks container status, verifies debugpy installation, auto-discovers target PID if not provided, checks if port is already listening, builds and executes the debugpy attach command via docker exec, and returns a structured result with status and next steps.
    @mcp.tool()
    def debugpy_attach(container: str, pid: int | None = None, port: int = DEFAULT_PORT, host: str = DEFAULT_HOST, python_bin: str = "python", wait_for_client: bool = False, log_to: str | None = DEFAULT_DEBUGPY_LOG_DIR, configure_subprocess: bool = False) -> dict[str, Any]:
        notes: list[str] = []
        if not docker_inspect_running(container):
            return DebugpyAttachResult(ok=False, container=container, port=port, host=host, notes=["Container is not running or does not exist."]).model_dump()
        installed, version = detect_debugpy_installed(container, python_bin=python_bin)
        if not installed:
            return DebugpyAttachResult(ok=False, container=container, port=port, host=host, notes=["debugpy is not installed inside the container.", "Install it in the image or running container before attach."]).model_dump()
        if version:
            notes.append(f"Detected debugpy {version} inside the container.")
        processes = get_process_table(container)
        if pid is None:
            pid, pid_notes = choose_pid(processes)
            notes.extend(pid_notes)
        if pid is None:
            return DebugpyAttachResult(ok=False, container=container, port=port, host=host, notes=["No candidate Python process was found to attach to."]).model_dump()
        if port_is_listening(container, port):
            mapped = docker_port_mapping(container, port)
            return DebugpyAttachResult(ok=True, container=container, port=port, host=host, pid=pid, already_listening=True, attached=False, mapped_port=mapped, notes=[f"Port {port} is already listening inside the container.", "Skipping injection because debugpy likely already attached."], next_steps=["Start your existing Attach configuration in Cursor.", "Verify your pathMappings match the container source path."]).model_dump()
        if log_to:
            docker_exec(container, f"mkdir -p {shlex.quote(log_to)}", timeout=10, check=False)
        cmd = build_debugpy_attach_cmd(python_bin=python_bin, host=host, port=port, pid=pid, wait_for_client=wait_for_client, log_to=log_to, configure_subprocess=configure_subprocess)
        proc = docker_exec(container, cmd, timeout=60, check=False)
        mapped_port = docker_port_mapping(container, port)
        listening = port_is_listening(container, port)
        if not listening:
            notes.append("debugpy did not appear to open the listening port after attach.")
            notes.append("Common causes: ptrace restrictions, wrong PID, or missing process privileges.")
            if "operation not permitted" in (proc.stderr or "").lower():
                notes.append("The container likely lacks ptrace permission for PID attach.")
            next_steps = [
                "Inspect stderr from the command output.",
                "Check container capabilities such as SYS_PTRACE and seccomp settings.",
                "Verify you attached to the worker process rather than a supervisor or master process.",
            ]
        else:
            notes.append(f"debugpy is now listening on {host}:{port} inside the container.")
            next_steps = [
                "Start your existing Attach configuration in Cursor.",
                "Set breakpoints in the relevant FastAPI route, dependency, or middleware path.",
                "If breakpoints do not bind, verify localRoot and remoteRoot path mappings.",
            ]
        return DebugpyAttachResult(ok=listening, container=container, port=port, host=host, pid=pid, attached=listening, mapped_port=mapped_port, command=cmd, stdout=(proc.stdout or "").strip() or None, stderr=(proc.stderr or "").strip() or None, notes=notes, next_steps=next_steps).model_dump()
  • Pydantic model DebugpyAttachResult that defines the output schema for the debugpy_attach tool, including fields for ok status, container, port, host, pid, already_listening, attached, mapped_port, command, stdout, stderr, notes, and next_steps.
    class DebugpyAttachResult(BaseModel):
        ok: bool
        container: str
        port: int
        host: str
        pid: int | None = None
        already_listening: bool = False
        attached: bool = False
        mapped_port: str | None = None
        command: str | None = None
        stdout: str | None = None
        stderr: str | None = None
        notes: list[str] = Field(default_factory=list)
        next_steps: list[str] = Field(default_factory=list)
  • Helper function build_debugpy_attach_cmd that constructs the shell command string for running debugpy in listen mode with the specified parameters (python binary, host, port, pid, wait_for_client, log_to, configure_subprocess).
    def build_debugpy_attach_cmd(*, python_bin: str, host: str, port: int, pid: int, wait_for_client: bool, log_to: str | None, configure_subprocess: bool) -> str:
        parts = [
            shlex.quote(python_bin), "-m", "debugpy", "--listen", f"{shlex.quote(host)}:{port}",
            "--configure-subProcess", "true" if configure_subprocess else "false"
        ]
        if wait_for_client:
            parts.append("--wait-for-client")
        if log_to:
            parts.extend(["--log-to", shlex.quote(log_to)])
        parts.extend(["--pid", str(pid)])
        return " ".join(parts)
Behavior1/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Tool has no description.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness1/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Tool has no description.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness1/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Tool has no description.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters1/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Tool has no description.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose1/5

Does the description clearly state what the tool does and how it differs from similar tools?

Tool has no description.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines1/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Tool has no description.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/will-garrett/debugpy-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server