--- title: 'Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hard' link: https://github.com/guregodevo/memdoor/tree/main/examples/workflows/team-solve source: hnrss-org published: 2026-10-31T22:46:12Z updated: 2026-10-31T22:46:12Z first_seen: 2026-11-01T09:08:30.968650873Z authors: - guregodevo content: extracted html: 2026-10-31-memdoor-a-team-of-5-qwen3-8-27b-models-beats-fable-5-on.html remote_preview: url: https://opengraph.githubassets.com/31e7563ac084244a417ccae15e55c468113b8911a2d9fea5a0472d4ebd45d06e/guregodevo/memdoor alt: A lightweight coding and automation agent in one CLI binary. Run locally or over SSH, optimize token use, and automate CI and cron with durable workflows. - guregodevo/memdoor --- Five small models solve a programming problem together: a planner, two ideators, a solver and a manager, plus a repair worker. Each starts from a fresh context and talks only through files on disk. A script that runs the samples decides success, not a model. It is the method of [GVS5H](https://github.com/slee-persis/GVS5H) (Lee, 2026), written as a [Memdoor](https://github.com/guregodevo/memdoor/blob/main/README.md) workflow: eight YAML files, no code, no model named. ## The result The authors' harness rebuilt on the same model, Qwen3.8-27B, with their prompts, loop and 128K reply cap. The first ten problems of their LiveCodeBench Hard list, five passes each, graded on every hidden test with their corrected grader: | pass | this workflow | authors' harness | Fable 5, one call | | ---- | ------------- | ---------------- | ----------------- | | 1 | **10** / 10 | 9 | 7 | | 2 | **10** / 10 | 10 | 8 | | 3 | **10** / 10 | 10 | 10 | | 4 | **10** / 10 | 9 | 8 | | 5 | **9** / 10 | 10 | 7 | | mean | **9.8** | 9.6 | 8.0 | The authors' two columns are the regraded result files in their repository, the numbers in their paper. Our column is [`bench/results/parity-5pass/`](https://github.com/guregodevo/memdoor/blob/main/examples/workflows/team-solve/bench/results/parity-5pass): all 50 runs, each with every request and reply, the program, the grader's output and the cost. A run took a median of 40 minutes and $0.39; the 50 cost $35.79. What this does and does not show: - Ten problems, five passes, not the authors' 100. On a sample this small, 9.8 and 9.6 are the same number. The claim is parity with their harness, not a gain on it. - The contests date from spring 2025 and the model may have seen them in training. Contamination cannot be ruled out, for any column. - Fable 5's column is one call per problem, from the authors' files. It was not run in this harness. - The loop sees only the public samples; the score is the hidden tests. - The model ran on OpenRouter, not a local machine. - The one lost problem: the manager invented a wrong counterexample and the team rewrote a correct program. Twice elsewhere a correct program was overwritten and then recovered. Both gaps are the authors' loop, kept as is. The parity check (every request byte for byte against their code), the hosts, the changes it took and the first run's findings are in [NOTES.md](https://github.com/guregodevo/memdoor/blob/main/examples/workflows/team-solve/NOTES.md). ## Run it ``` curl -fsSL https://memdoor.ai/install.sh | bash memdoor connect openrouter # or anthropic, openai, gemini, chatgpt # or no key: memdoor connect local (Ollama, llama.cpp, vLLM, LM Studio; Ollama needs a 32K window, not its 4K default) git clone --depth 1 https://github.com/guregodevo/memdoor mkdir -p ~/.memdoor/workflows cp -r memdoor/examples/workflows/team-solve ~/.memdoor/workflows/ mkdir my-problem && cd my-problem cp -r ../memdoor/examples/workflows/team-solve/sample/* . # task.md + tests/1.in … 5.out memdoor workflow run team-solve --wait # exit 0 when every sample passes ``` For your own problem, replace `task.md` and put paired `tests/N.in` / `tests/N.out` beside it. The program is `solution.py`, read from stdin, compared to the samples with `diff`; it needs Python 3 and bash. `/workflow` in `memdoor tui` draws the run live. The steps name no model: whatever `/model` is pinned to answers, a local one included ([how to connect one](https://memdoor.ai/docs/providers#a-local-runtime-no-key)). To change a role, change its prompt in `steps/`. ## How it works ``` plan ──┬── ideate-a ──┬── solve ── test ── review ── repair ── verify └── ideate-b ──┘ ``` | step | kind | reads | writes | | ---------------------- | ------------------ | ---------------- | -------------------------------------------------------------------------------- | | `plan` | agent | `task.md` | `plan.md`, `notes.md` | | `ideate-a`, `ideate-b` | agent, in parallel | the plan | a proposal each: the most direct provable algorithm, and one from another family | | `solve` | agent | both proposals | `solution.py`, the decision in `notes.md` | | `test` | command | `tests/` | `test-output.log` (never fails the run) | | `review` | agent | the test output | repair instructions in `notes.md` | | `repair` | agent | the instructions | a fixed `solution.py`, or nothing | | `verify` | command | `tests/` | the verdict: a nonzero exit fails the run | Every agent prompt says "fresh context, read the files, do not rely on chat". The paper credits that ledger discipline for the gain. Each step is one YAML file in `steps/`, in the format of [mario](https://github.com/guregodevo/mario), the engine underneath; `requires:` is the DAG.