Skip to main content
Glama
bayleafwalker

hostproto-dap-debugpy

hostproto-dap-debugpy

The HostProto DAP adapter (hostproto-semantics plan, step 5): a debugpy session exposed through MCP 2026-07-28, pinned. Semantics come from hostproto-semantics, the same eleven bundles the browser adapter pins; this repository owns only the DAP client, the thread/target registry, validation, and its own tests. It is the runtime the step-3 spike (docs/DAP_SPIKE.md) asked for. Since hostproto-semantics ADR-0012 it is a binding: src/binding.ts supplies how debugpy is started and what its launch says; hostproto-dap-core (pinned by commit) computes every HostProto semantic and serves MCP.

What it does

MCP surface

HostProto object

hostproto_context_create (launch a Python program, stops on entry)

handles/v1 — host = adapter process, context = launch, surface = thread; lifecycle: creating until the first thread exists

hostproto_surface_observe

observation/v1 — revision moves on every stopped↔running transition of that thread; the cursor is host-assigned over normalized events, DAP seq is raw provenance. Projections: state, frames, scopes, variables, output, breakpoints, host_requests

hostproto_surface_act (input is the intent/v1 bundle)

receipt/v1set_breakpoints, step_over/in/out, continue, pause, evaluate, set_variable, host_request.resolve; a resume whose deadline elapses is outcome: unknown, executed: false

hostproto_surface_await

host-side wait: stopped, lifecycle, revision, event_kind, host_request

hostproto_context_recovery

recovery/v1unrecoverable / host_terminated once the debuggee is gone, with the raw DAP message log as content-addressed evidence (application/x-ndjson)

hostproto_capabilities

capability-profile/v1 for dap/v1 — availability from the adapter's initialize response, runtime only for what this process executed

hostproto://surface/{id}/state, hostproto://context/{id}/dap-messages

subscribable state; the evidence log

Frames, scopes and variables are target-ref/v1 scoped to the revision they were observed at. A variablesReference from before a resume is refused as target_invalidated with host_invoked: false before anything is sent — the rule step 3 asked step 5 to prove on a real debugger. Frames while running are omitted, not lost: bounded.omitted counts them, lossy stays false, a deviation says why.

A runInTerminal reverse request is a host request behind a decision token, the same shape as a browser script dialog: handles come back first (creating), the client observes host_requests, resolves with allow | deny, and the entry stop follows.

Related MCP server: vibe-debug

Run

npm ci
python3 -m venv .venv && .venv/bin/pip install debugpy==1.8.21   # or HOSTPROTO_PYTHON=/path/to/python with debugpy
npm test          # real client ↔ real server over stdio ↔ real debugpy ↔ real Python
npm start         # stdio server
A
license - permissive license
Not graded
quality - not tested
B
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

  • A
    license
    A
    quality
    D
    maintenance
    Enables AI assistants to perform interactive Python debugging with breakpoints, step execution, and variable inspection using the Debug Adapter Protocol (DAP) through an MCP server interface.
    8
    1
    MIT
  • A
    license
    Not graded
    quality
    D
    maintenance
    Enables coding agents to use a real debugger (Python via debugpy) for launching, attaching, setting breakpoints, stepping through code, inspecting stack frames, and evaluating expressions through MCP tools.
    27
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI agents to inspect debug state, control execution, and set breakpoints in VS Code by exposing the Debug Adapter Protocol as an MCP server.
    Apache 2.0
  • A
    license
    Not graded
    quality
    B
    maintenance
    Exposes VS Code's Debug Adapter Protocol as MCP tools, enabling AI coding agents to set breakpoints, step through code, inspect variables, and read exceptions in a real debug session.
    GPL 3.0

View all related MCP servers

Related MCP Connectors

  • Live browser debugging for AI assistants — DOM, console, network via MCP.

  • Agent Replay Debugger MCP — record every agent step + deterministic replay. Step-debugger for

  • MCP server exposing the Backtest360 engine API as tools for AI agents.

View all MCP Connectors

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/bayleafwalker/hostproto-dap-debugpy'

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