debuff

jp.debuffPublished by debuff.jpv0.32.0Updated Nov 7, 2026

Let coding agents play, test and tune your 2D web game: live state, explained errors, replays, fuzz.

VerifiedSTDIODesktop onlyBrowser AutomationDeveloper ToolsMedia & Design

Overview

debuff's local MCP server, published by debuff.jp; no sign-in required; STDIO.

No AI overview yet: these facts come from the registry entry and the server itself.

Publisher
debuff.jp

Installation

In SourceWeft

  1. Open debuff in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

@debuffjp/mcp

An MCP server that lets AI agents inspect, build, test and play a debuff game. It uses the same evidence a human would want: real state, real frames, real test results.

Setup

New projects from npm create debuff include a .mcp.json, so MCP clients that read it (such as Claude Code) pick the server up automatically. For other clients:

bash
claude mcp add debuff -- npx -y @debuffjp/mcp        # Claude Code
json
{  "mcpServers": {    "debuff": { "command": "npx", "args": ["-y", "@debuffjp/mcp"] }  }}

Run it from the game's folder, or pass --root <dir> (or set DEBUFF_ROOT).

As a Claude Code plugin, which also adds a skill that teaches Claude the workflow:

bash
/plugin marketplace add debuffjp/debuff/plugin install debuff@debuff

As a Codex plugin: codex plugin marketplace add debuffjp/debuff, then codex plugin add debuff@debuff.

For Claude Desktop, mcpb/manifest.json describes a desktop extension that asks for the game folder. Pack it from that folder with npx @anthropic-ai/mcpb pack.

Tools

ToolWhat it does
inspect_projectDependencies, plugins and their contributions, assets, last build size
list_assetsEvery asset with its type, size and importer, plus warnings and possibly unused ids
typecheckType-check with the project's own tsc; no browser needed. Vite doesn't type-check, so run it after edits
buildProduction build plus size report; leaks must be empty
run_testsRun the logic tests in tests/ (keiko): the real game in Node, no browser; failures carry the tick, scene and state
run_scenariosRun scenarios/*.json headless; per-step results, errors, captures
fuzzPlay with random input; each crash, stuck state or back-and-forth loop comes with an explained error and a scenario that replays it
game_startAttach to the live game, or open a private headless copy, paused (see below). Headless takes a viewport: [w, h]; reducedMotion forces it on or off
game_stepAdvance fixed simulation steps (pausing first); returns state (quiet leaves out your inspect() data)
game_pausePause or resume a live game
game_framesRecent frame breakdown: update/render ms, catch-up steps, quads, draw calls, long tasks; summary gives percentiles and the 3 slowest
game_notesMoments the human flagged in the overlay, with the state at that tick
game_rewindLive mode: take the human's game back (or forward) to a tick, paused, with its exact state then
game_replaySave the live session up to a tick (for example a note's) as a scenario in .debuff/replays/ that reproduces it. From 0.31 it holds state checkpoints, so run_scenarios warns at the tick where the replay stops matching
game_inputpress / release keys (and PointerRight / PointerMiddle); tap, pointer_down (with an optional kind: mouse, touch or pen), pointer_move, pointer_up; wheel (dy, dx); touch_down, touch_move, touch_up by contact id
game_stateScene, tick, logical width/height, renderer, frame stats, reducedMotion, your inspect() data
game_commandRun a command the game registered with devTools({ commands }), with JSON args
game_params / game_set_paramRead and tune typed params (validated)
game_captureThe current frame as a PNG image; path also saves it in the project (thumbnails, store art). social: true letterboxes it into 1200×675 for X and Bluesky posts (added in 0.26)
game_clipLive mode: start a rolling recording, save the last 15 s as a video in .debuff/captures/, stop. For devlogs and press kits; the human posts. Added in 0.26
game_errorsRuntime errors, explained: whose problem (your code, a rejected call, a debuff bug), the source line, a hint
game_stopClose the game. From 0.28, a headless game also closes after 15 idle minutes, and a call that takes over 60 s fails with game_hung
report_papercutReport tooling friction in debuff, keiko, myaku, spryte or debuff.jp to papercuts.debuff.jp. Returns close matches instead of filing a duplicate; call again with hit: <id> or force: true. With hit, workaround fills the papercut's workaround when it has none (from 0.24.1). Added in 0.24

Tool calls that fail on setup twice in a week, or run far too long, are reported to papercuts.debuff.jp on their own, and the result says so (see debuff papercuts in the CLI README to turn it off). From 0.24.

Prompts

Workflows clients can offer as slash commands: playtest, fix-errors, tune-feel, add-mechanic, devlog-post (added in 0.26: capture a still and a clip, then draft short posts for the human to send) and explain-stutter.

Live and headless

If you have the game open under debuff dev, game_start attaches to that tab (mode live). The overlay shows an agent badge, and every pause, step and param change the agent makes appears in its activity log, where you can undo it. Type a note in the overlay ("jump feels floaty here") to flag a moment; the agent reads it, along with the game state at that tick, through game_notes. game_stop resumes your game and disconnects.

Otherwise, or with mode: "headless", the agent gets a private copy in a headless browser. Time only advances through game_step, so what it sees is reproducible, and a scenario can replay it exactly.

Source: tooling/mcp/README.md at commit 247493d

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

24
  1. v0.32.0LatestNov 7, 2026
  2. v0.31.0Nov 6, 2026
  3. v0.29.0Nov 3, 2026
  4. v0.27.0Nov 1, 2026
  5. v0.26.2Nov 1, 2026
  6. v0.26.0Oct 31, 2026
  7. v0.24.3Oct 31, 2026
  8. v0.24.2Oct 31, 2026
  9. v0.24.0Oct 30, 2026
  10. v0.23.0Oct 30, 2026
  11. v0.21.1Oct 30, 2026
  12. v0.15.0Oct 23, 2026
  13. v0.14.0Oct 23, 2026
  14. v0.13.1Oct 22, 2026
  15. v0.13.0Oct 21, 2026
  16. v0.11.0Oct 21, 2026
  17. v0.9.5Oct 20, 2026
  18. v0.9.4Oct 20, 2026
  19. v0.9.2Oct 19, 2026
  20. v0.9.0Oct 19, 2026
  21. v0.7.1Oct 18, 2026
  22. v0.7.0Oct 18, 2026
  23. v0.6.0Oct 18, 2026
  24. v0.5.0Oct 18, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts