Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
[ICML 2026] A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation
| Date | Stars |
|---|---|
| 2026-07-31 | 678 |
| 2026-08-05 | 701 |
| 2026-08-06 | 701 |
| 2026-08-14 | 730 |
| 2026-08-18 | 742 |
| 2026-08-19 | 741 |
| 2026-08-20 | 747 |
| 2026-08-21 | 749 |
| 2026-08-22 | 748 |
| 2026-08-23 | 749 |
| 2026-08-24 | 754 |
| 2026-08-25 | 756 |
| 2026-08-26 | 758 |
| 2026-08-27 | 760 |
| 2026-08-29 | 761 |
| 2026-08-30 | 762 |
| 2026-08-31 | 766 |
| 2026-09-02 | 767 |
| 2026-09-03 | 769 |
| 2026-09-06 | 770 |
| 2026-09-07 | 775 |
| 2026-09-08 | 779 |
| 2026-09-10 | 782 |
| 2026-09-11 | 785 |
| 2026-09-12 | 788 |
| 2026-09-13 | 789 |
| 2026-09-14 | 792 |
| 2026-09-15 | 797 |
| 2026-09-16 | 802 |
| 2026-09-17 | 805 |
| 2026-09-18 | 808 |
| 2026-09-20 | 813 |
Today
+5 stars today
This week
+24 stars this week
This month
+64 stars this month
Momentum
0.0
growth rate 3.04%/day
# CaP-X
### A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation
[Project Page](https://capgym.github.io/)  |  [Paper](https://arxiv.org/abs/2603.22435)
**Max Fu<sup>*,1,2</sup>, Justin Yu<sup>*,2</sup>, Karim El-Refai<sup>*,2</sup>, Ethan Kou<sup>*,2</sup>, Haoru Xue<sup>*,1,2</sup>,
Huang Huang<sup>3</sup>, Wenli Xiao<sup>4</sup>, Guanzhi Wang<sup>1</sup>, Fei-Fei Li<sup>3</sup>, Guanya Shi<sup>4</sup>, Jiajun Wu<sup>3</sup>,
Shankar Sastry<sup>2</sup>, Yuke Zhu<sup>1</sup>, Ken Goldberg<sup>†,2</sup>, Jim Fan<sup>†,1</sup>**
<sup>1</sup>NVIDIA   <sup>2</sup>UC Berkeley   <sup>3</sup>Stanford University   <sup>4</sup>Carnegie Mellon University
<sup>*</sup>Equal contribution   <sup>†</sup>Equal advising
---
**CaP-X** is an open-access framework for systematically studying Code-as-Policy agents in robot manipulation. It consists of four components:
| Component | What it does |
| -------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **CaP-Gym** | Interactive Gymnasium environments where agents control robots by generating Python code that composes perception and control primitives. 39 tasks across Robosuite, LIBERO-PRO, and BEHAVIOR. |
| **CaP-Bench** | Systematic benchmark evaluating coding agents across abstraction levels, interaction modes, and visual grounding modalities. 8 tiers (S1-S4 single-turn, M1-M4 multi-turn). |
| **CaP-Agent0** | Training-free agentic framework with multi-turn visual differencing, auto-synthesized skill libraries, and parallel ensembled reasoning. |
| **CaP-RL** | Reinforcement learning on the coding agent via GRPO, using environment rewards to post-train language models. Transfers from sim to real with minimal gap. |
---
## Installation
CaP-X uses [uv](https://docs.astral.sh/uv/) for dependency management. Requires **Python 3.10** and a **CUDA-capable GPU**.
```bash
git clone --recurse-submodules https://github.com/capgym/cap-x && cd cap-x
# Or if already cloned without --recurse-submodules:
git submodule update --init --recursive
# Install uv (if not present)
curl -LsSf https://astral.sh/uv/install.sh | sh
uv python install 3.10 && uv venv -p 3.10
# Base install
uv sync
```
### Simulator-specific setup
Pick **one** simulator family to install, as Robosuite (1.5.0) and LIBERO (`robosuite==1.4.0`) would be in conflict.
#### Robosuite
```bash
uv sync --extra robosuite
```
#### LIBERO-PRO
LIBERO requires a **separate virtual environment**.
```bash
uv venv .venv-libero --python 3.12
source .venv-libero/bin/activate
uv sync --active --extra libero --extra contactgraspnet
```
See [docs/libero-tasks.md](docs/libero-tasks.md) for running any of 130+ LIBERO tasks.
#### BEHAVIOR (Isaac Sim)
BEHAVIOR tasks run on NVIDIA Isaac Sim via OmniGibson. Requires Python 3.10 and CUDA 12.x.
```bash
cd capx/third_party/b1k
./uv_install.sh --dataset # installs OmniGibson, Isaac Sim, BDDL, cuRobo, and downloads assets
cd ../../.. # back to repo root
# Post-install fix — copy cuRobo JIT headers (run with b1k venv active)
source capx/third_party/b1k/.venv/bin/activate
cp capx/third_party/curobo/src/curobo/curobolib/cpp/*.h \
$(python -c "import sysconfig; print(sysconfig.get_path('purelib'))")/curobo/curobolib/cpp/
```
> The `--dataset` flag downloads robot assets, BEHAVIOR-1K scene/object assets, and 2025 challenge task instances. You will be prompted to accept the NVIDIA Isaac Sim EULA and BEHAVIOR datExcerpt of 9,517 characters
Read on GitHub
Claude · @anthropics
3
2
2
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:3e16ddf8ada7aebf, desc:manipulation