Menu

Individual skill

setup-boltzgen-remote-tool

0upvotes

Set up, launch, validate, and troubleshoot the BoltzGen ToolUniverse remote tool and optionally relay it through ToolUniverse Connect. Use when deploying or auditing this implementation.

skills/setup-boltzgen-remote-tool/SKILL.md

Read pinned source

Source commit 5598a9fc97e32964fe9b703818e7d0747142f179

What was checked

Matches the observed repository HEAD
Yes
Scanner rules are current
Yes
Repository scan complete
No
Skill scan complete
No
Observed behavior
Network or broad shell
Last checked
Nov 9, 2026

These are observations at the recorded commit, not a guarantee of safety. Review the source before running it.

Install this skill

Choose an agent in your Stack or copy a command below. Commands fetch the pinned commit and copy only this skill directory. Existing destinations are refused. Copying a command does not install anything.

An install command is unavailable: current rules, complete repository and skill scans, an eligible behavior level and a permitted license are required.

Repository context

Repository stars (not skill ratings)
1,780
Repository last updated
Nov 8, 2026
License
Apache-2.0

Observed agent integration: No specific integration evidence. Standard SKILL.md installation does not imply verified compatibility.

Files and excerpt

File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.

Browse files
Read excerpt

# Set up BoltzGen as a remote tool

> Validation status (2026-10-01): BoltzGen 0.3.2 ran on the NVIDIA GB10 (aarch64, cc 12.1) under the optimization kit's `fast` mode with all eight in-process levers. Verified on FOUR CONSECUTIVE IDENTICAL REQUESTS in one held worker on an idle card -- 27.4 / 26.6 / 27.2 / 29.1 s, peak reserved 2.49 GiB on all four -- plus a seven-shape mixed sequence varying length, design count, batch size, step count and step list under one process. **`[kit.env] OMP_NUM_THREADS = "1"` is load-bearing and must not be removed.** Without it the design step does not finish: upstream's DataLoader forks a worker that enters `torch.cdist` -> `at::parallel_for` -> an OpenMP parallel region, and waits forever on a thread pool `fork` did not carry over. The kit's levers do host-tensor work during activation, so the pool is already up when the fork happens; stock upstream has not touched it yet, which is why stock computes at 95% GPU and the kit mode sits at 0%. Measured: unset = 0 designs after 420 s; `=1` = 31.6 s. Peak scales with chain length (3.17x over 10x) and barely with design count (1.06x from 2 to 8 designs), so budget this one by length. Public publication, cross-user isolation, concurrency, recovery and design quality are untested; keep this deployment private.

## Prerequisites

- Run from the ToolUniverse repository root on Linux with Python 3.12.3.
- A CUDA GPU is required. This provider holds its model resident between requests; CPU is not a target.
- Keep provider data, weights, caches, and credentials outside Git.
- Bind to loopback. A non-loopback bind requires TOOLUNIVERSE_API_TOKEN; never put it in arguments or results.

Run the standard-library contract check before downloading large dependencies:

~~~bash
Back to top