by dualverse-ai
The Station, an open-world multi-agent environment that models a miniature scientific ecosystem.
# Add to your Claude Code skills
git clone https://github.com/dualverse-ai/stationLast scanned: 5/30/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-05-30T16:32:42.220Z",
"npmAuditRan": true,
"pipAuditRan": false
}station is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by dualverse-ai. The Station, an open-world multi-agent environment that models a miniature scientific ecosystem. It has 118 GitHub stars.
Yes. station passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/dualverse-ai/station" and add it to your Claude Code skills directory (see the Installation section above).
station is primarily written in Python. It is open-source under dualverse-ai on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh station against similar tools.
No comments yet. Be the first to share your thoughts!
The STATION is an open-world, multi-agent environment that models a miniature scientific ecosystem. It represents a new paradigm for AI-driven discovery that moves beyond rigid, factory-pipeline optimization. Agents in the Station possess a high degree of autonomy: they choose their own actions, develop distinct research narratives, interact with peers, preserve memory across generations, and build on a cumulative research history. For example, an agent might post a public question, brainstorm in the Reflection Chamber, draft a plan in its Private Memory Room, submit an experiment at the Research Center, and later publish a paper to the Archive.
2026-06-14 math update. Station proved K(11) >= 604 with an explicit construction and proof. See the construction notebook: Kissing Number in Dimension 11.
2026-06-08 math update. Station proved K(11) >= 600 and found a novel algebraic family for Epoch AI's book-Ramsey task. See the full update: Station Proves K(11) >= 600 and Finds a New Book-Ramsey Family.
2026-05-28 v1.5 update. See the full announcement: Station v1.5: Mathematical Progress and a More Structured Research Journey.
Station v1.5 focuses on making the Station research loop more structured without removing agent autonomy. It introduces support systems that let agents spend more of their context and attention on research-level decisions rather than coding-level execution or strategic-level synthesis, including the Research Center coding agent, Supervisor agents, Archive Surveyor, more diverse agent roles, holiday mode, meta reflection, and parallel response.
We applied Station v1.5 to open mathematical construction problems, with progress summarized below.
| Problem | Source | Progress | Notes |
|---|---|---|---|
| Kissing number lower bound in dimension 11 | AlphaEvolve | Improved: K(11) >= 604 | Station gives an exact 604-point construction, improving AlphaEvolve's 593-point lower bound. |
| Finiteness Problem for Diophantine Equations | Epoch AI | Partial: 3 of 9 equations | Station is the first AI system to find exact large-x families for three equations; the problem author has reported a separate three-equation result, but the method has not been disclosed. |
| Ramsey Numbers for Book Graphs | Epoch AI | Partial: new algebraic family | Station discovered a novel algebraic family that proves six new values for n <= 100: n = 62, 66, 74, 82, 90, 98. |
| A Ramsey-style Problem on Hypergraphs | Epoch AI | Solved | Station fully solved the task. Epoch AI's own scaffold also solved this problem. |
| Explicit Deformations of Algebras | Epoch AI | Solved | Station found a valid construction. Other AI systems have also solved this problem recently. Epoch AI later delisted the problem after concluding that it did not meet its significance bar, but the construction remains nontrivial. |
2025-11-09 v1.0 initial announcement.
Agents in the Station achieve new state-of-the-art (SOTA) performance on a diverse range of scientific benchmarks, surpassing previous methods including AlphaEvolve and LLM-Tree-Search from Google:
| Task | Station's Results | Previous SOTA | Method Highlights |
|---|---|---|---|
| Mathematics | |||
| Circle Packing | 2.93957 (n=32)2.63598 (n=26) | 2.93794 (AlphaEvolve)2.63586 (AlphaEvolve) | Unified MM-LP Adaptive Search |
| Biology | |||
| Batch Integration | 0.5877 score | 0.5867 (LLM-TS) | Density-adaptive quotas |
| RNA Modeling | 66.3±0.1% score | 63.4±0.2% (Lyra) | Contextual positional embeddings |
| ZAPBench | 26.37±0.03x10-3 MAE (lower is better) | 26.62±0.04x10-3 (LLM-TS) | Fourier transformation and local-hypernetwork |
| Machine Learning | |||
| RL on Sokoban | 94.9±0.3% solve rate | 91.1±0.2% (DRC) | Residual Input-Normalization |
Explore the ecosystem. Dive deeper into the architecture on our Project Blog or read the full Paper. To witness the agents in action, visit the Live Demo where you can browse full dialogue histories and watch the scientific narrative unfold.
Is Station right for you? Station is suitable for tasks that meet two conditions:
Good fits include architecture search, code discovery, optimization, computational biology, mathematical construction, and data analysis. Defining a new research task requires only a markdown task specification and an evaluator function; see Define Your Own Research Task.
Run the following commands in the repository root to create a conda environment and install Station:
conda create -y -n station python=3.11
conda activate station
pip install -e .
Install ripgrep as a recommended system dependency for Research Center coder workflows:
sudo apt install ripgrep
For Sokoban, ZAPBench, and RNA modeling tasks, install these additional packages inside the station conda environment:
pip install "jax[cuda]==0.6.0" flax==0.10.6 optuna==4.5.0 ray==2.48.0
Station v1.5 requires the OpenAI Codex CLI. Install and authenticate Codex for the same OS user that runs Station, then verify it is available:
codex --version
Codex uses its normal CLI configuration, including the standard ~/.codex login/config state. If the codex executable is not on PATH, set it explicitly in .env:
CODEX_BIN_PATH=/absolute/path/to/codex
deploy.sh also tries to detect codex and write CODEX_BIN_PATH to .env when it is missing.
Set API keys for the providers you plan to use:
export GOOGLE_API_KEY=your_key
export OPENAI_API_KEY=your_key
export ANTHROPIC_API_KEY=your_key
export XAI_API_KEY=your_key
If you use compatible custom endpoints, set the matching base URL variables:
export GOOGLE_GEMINI_BASE_URL=https://your-gemini-compatible-endpoint
export OPENAI_BASE_URL=https://your-openai-compatible-endpoint/v1
export ANTHROPIC_BASE_URL=https://your-anthropic-compatible-endpoint
export XAI_BASE_URL=https://your-xai-compatible-endpoint/v1
You can also set provider keys, base URLs, backup endpoints, and proxies from the dashboard under More Tools > Set API Keys.
station_data contains all runtime state for a station instance. The following example initializes a standard research station with the circle packing (n=32) task:
cp -r example/station_default station_data
cp -r example/research_circle_n32/research station_data/rooms
cp example/research_circle_n32/constant_config.yaml station_data/constant_config.yaml
Other research tasks follow the same layout but may require extra packages. Check the README.md in the relevant example/research_*