File size: 6,334 Bytes
de1f371 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 | ---
license: apache-2.0
pretty_name: Tiled computer-use environment from a one-line request (REA)
tags:
- computer-use
- cua
- gui-agent
- rl-environment
- agentenv
- desktop
- linux
- rea
- game-development
- tile-map
- level-editor
- from-request
---
# Tiled: a computer-use environment built from a one-line request
**The real Tiled 1.8 map editor running on an Ubuntu 24.04 desktop, as an [agentenv](https://pypi.org/project/agentenv-framework/) computer environment. An agent completes the task only through screenshots, mouse and keyboard, and a verifier grades the saved files with no model.**
Claude Code built this environment from a one-line request in about 26 minutes, with only REA's build skill and MCP tools; [How it was made](#how-it-was-made) has the request and the steps. Tiled opens a small map, `Documents/harbor_town.tmx`, at launch: a 16 by 12 grid with an empty tile layer and a Landmarks object layer holding four objects, Old Well, Blacksmith Door, Harbor Gate and Player Spawn. The agent has to rename Old Well to Wishing Well and save the map. It sees only the screen and acts only with the mouse and keyboard.
## Watch
- Agent runs (Claude Code with Claude Sonnet 5.5): [run 1](media/solve-1.mp4), [run 2](media/solve-2.mp4), [run 3](media/solve-3.mp4)
- Witness, which proves the task can be done through the computer: [witness.mp4](media/witness.mp4)
Each video is the computer's own screen recording at 10 frames per second.
## The task
> In my Tiled map Documents/harbor_town.tmx, rename the object called Old Well to Wishing Well and save the map.
`bundle/artifacts/rename-old-well-check/verify.py` reads the app's files through the computer's `data/get` view and passes only if every check holds:
| Check | Kind | File |
| --- | --- | --- |
| `renamed` | `value_equals` | `Documents/harbor_town.tmx` |
| `old-name-gone` | `value_absent` | `Documents/harbor_town.tmx` |
| `door-kept` | `value_equals` | `Documents/harbor_town.tmx` |
| `gate-kept` | `value_equals` | `Documents/harbor_town.tmx` |
| `four-objects` | `member_count` | `Documents/harbor_town.tmx` |
A `.tmx` file is XML. The checks find objects by their `id` in the Landmarks object layer (`objectgroup` id 2): object 1 must be named "Wishing Well", no object may still be named "Old Well", objects 2 and 3 must keep their names, Blacksmith Door and Harbor Gate, and the layer must hold four objects (`member_count`). REA's draft suggested the first two checks, both kept exactly as written. It also suggested two guards that picked each neighbour by the very name they checked, which proves only that the name still exists somewhere; Claude Code switched them to the objects' ids and added the count.
## Results
3 of 3 runs passed, about $0.72 per run.
| Run | Outcome | Turns | Cost | Video |
| --- | --- | --- | --- | --- |
| 1 | passed | 9 | $0.70 | [video](media/solve-1.mp4) |
| 2 | passed | 13 | $0.74 | [video](media/solve-2.mp4) |
| 3 | passed | 11 | $0.71 | [video](media/solve-3.mp4) |
## Run it
You need Docker. `agentenv-framework` builds the computer: Ubuntu 24.04 with the app installed from Ubuntu's archive (`tiled` 1.8.2).
```bash
pip install agentenv-framework==0.9.1431
hf download earakely-scale/TiledFromRequestComputerEnv-REA --repo-type dataset --local-dir tiled-request
agent-env run tiled-request/bundle --task rename-old-well --agent claude-code --model <model>
```
The agent needs a model provider configured for agentenv. `--task smoke` only boots the computer, with no agent. `bundle/README.md` describes the bundle in detail.
## How it was made
Claude Code built this environment from a one-line request, given only [REA](https://github.com/morluto/rea)'s build skill and REA's MCP tools:
> Build me a Tiled environment where the agent renames an object in a small tile map and saves it.
It asked only the skill's upfront questions, wrote no app-specific scripts, and finished in about 26 minutes. The steps were REA's usual ones:
1. **Capture.** REA launched Tiled on a Linux desktop and recorded its package, windows and opened files while the edit was made. It also recorded the app's state folders before and after.
2. **Export.** REA turned the capture into an agentenv `computer_app`. It installs the app, seeds the starting files, starts the app, and waits for its main window. It also serves typed views of the app's files for verifiers.
3. **Task.** The task's checks were validated against the capture: they fail on the starting state and pass on the captured end state.
4. **Witness.** In a fresh computer, the object was renamed through computer actions alone, and the checks were read at four stages. They held on the first try:
| Stage | Checks |
| --- | --- |
| `seed` | fail, as expected: object 1 is still named Old Well; the kept checks pass |
| `after_edit` | pass |
| `near_miss` | fail, as expected: the object was renamed "Wishing Wel" |
| `after_reset` | fail, as expected: `data/reset` restored the seed |
The agent runs came after the build: Claude Code with Claude Sonnet 5.5 (`--model anthropic/claude-sonnet-5-5 --no-emulate`, on an amd64 host) solved the task 3 times, each run recorded.
## What's inside
```
bundle/ the agentenv bundle: envs/ (app module, seed files), tasks/, artifacts/ (verifier), evals/
media/ witness and agent-run recordings (MP4) with short GIF previews
results.json the runs above, with each check's result
```
## Limits
- The map has no tiles: its tile layer is empty and it has no tileset, so it shows only its four objects, as outlined shapes on a grid.
- The renamed object is found by its id, 1: a new object made in place of Old Well gets another id and fails. Only the Blacksmith Door and Harbor Gate names are checked as kept; Player Spawn's name and every object's position and size are not.
- The computer installs whatever `tiled` Ubuntu 24.04's archive holds at build time (1.8.2 for these runs); it is not pinned.
- Three runs is a small sample: the solve rate shows the task is solvable, not how hard it is.
## Licences
The bundle code, the task and the synthetic seed files are Apache-2.0. Tiled itself is not included: the computer installs it from Ubuntu's archive (`tiled` 1.8.2) at build time, under its own licence (GPL-2.0-or-later).
|