File size: 7,185 Bytes
e5dd13c | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 | ---
license: apache-2.0
pretty_name: Tellico computer-use environment from a one-line request (REA)
tags:
- computer-use
- cua
- gui-agent
- rl-environment
- agentenv
- desktop
- linux
- rea
- collection-manager
- catalog
- books
- from-request
---
# Tellico: a computer-use environment built from a one-line request
**The real Tellico 3.5 collection manager 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 28 minutes, with only REA's build skill and MCP tools; [How it was made](#how-it-was-made) has the request and the steps. Tellico opens a small book collection, `Documents/my-books.tc`, at launch: three synthetic books with their authors, publishers and years. The agent has to add a fourth, The Glass Orchard by Aurelio Benn, published by Northwind House in 2022, and save the collection. 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
> Add the book The Glass Orchard by Aurelio Benn to my book collection in Documents/my-books.tc, with Northwind House as the publisher and 2022 as the publication year, and save it.
`bundle/artifacts/add-glass-orchard-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 |
| --- | --- | --- |
| `title` | `text_equals` | `Documents/my-books.tc` |
| `author` | `text_equals` | `Documents/my-books.tc` |
| `publisher` | `text_equals` | `Documents/my-books.tc` |
| `pub-year` | `text_equals` | `Documents/my-books.tc` |
| `keeps-first-book` | `text_equals` | `Documents/my-books.tc` |
| `keeps-third-book` | `text_equals` | `Documents/my-books.tc` |
A Tellico `.tc` file is a zip holding `tellico.xml`, and REA reads the entry inside the zip. Each book is an `entry` element with an `id`, and the checks find books by it, never by position: the new book, entry 4, must have exactly the title, author, publisher and publication year asked for, and books 1 and 3 must keep their titles. REA's draft suggested one check on the whole collection's text, which takes in the date Tellico stamps on a new book and would fail on any other day; Claude Code discarded it and wrote these from REA's view of the added entry. The verifier grades once the collection file has held still for 3 seconds.
## Results
3 of 3 runs passed, about $0.90 per run.
| Run | Outcome | Turns | Cost | Video |
| --- | --- | --- | --- | --- |
| 1 | passed | 24 | $1.26 | [video](media/solve-1.mp4) |
| 2 | passed | 23 | $0.93 | [video](media/solve-2.mp4) |
| 3 | passed | 21 | $0.50 | [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 (`tellico` 3.5.3).
```bash
pip install agentenv-framework==0.9.1431
hf download earakely-scale/TellicoFromRequestComputerEnv-REA --repo-type dataset --local-dir tellico-request
agent-env run tellico-request/bundle --task add-glass-orchard --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 Tellico environment where the agent adds a book to a small book collection and saves it.
It asked only the skill's upfront questions, wrote no app-specific scripts, and finished in about 28 minutes.
It took a second capture. The first bundle's witness held, but Tellico showed its Tip of the Day on every boot: the setting that turns it off is kept in `~/.config/tellicorc`, which that capture did not own. Claude Code captured again naming that file as a state path, and this bundle's witness held on the first try too. The steps were REA's usual ones:
1. **Capture.** REA launched Tellico 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 book was added through computer actions alone, and the checks were read at four stages. They held on the first try:
| Stage | Checks |
| --- | --- |
| `seed` | fail, as expected: the collection has no entry 4; the kept-book checks pass |
| `after_edit` | pass |
| `near_miss` | fail, as expected: the author was retyped as "Aurelio Ben" |
| `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
- Adding a book takes Tellico's entry dialog across two tabs, title and author on General, publisher and year on Publishing: the runs took 21 to 24 turns.
- The checks find the new book as entry 4, the id Tellico gives the next book in this collection. A book that ends up with another id, for example one added after a first attempt was deleted, is not found and fails.
- Only the first and third books' titles are checked as kept, so editing Salt Roads, or the other fields of the first and third books, still passes.
- The seed also holds storage and GPU cache files of Tellico's built-in web view (`~/.local/share/tellico/QtWebEngine`), which the capture picked up. Tellico started cleanly from them in the witness and in every run.
- The computer installs whatever `tellico` Ubuntu 24.04's archive holds at build time (3.5.3 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. Tellico itself is not included: the computer installs it from Ubuntu's archive (`tellico` 3.5.3) at build time, under its own licence (GPL-2.0-only or GPL-3.0-only).
|