The progressive-disclosure recipe for Chai-1, an all-atom diffusion co-folder that treats protein, RNA, DNA, and SMILES-ligand chains as first-class entities in one multi-entity FASTA. It walks the agent through the external chai-lab Python API and through the ranked candidates and scores that come back; the model itself is not shipped here. Chai-1 covers much the same ground as boltz, and the recipe leans on that: running both and keeping the designs that pass either is a common consensus filter, and Chai's Python entry point makes it the easier of the two to embed in a design loop.
The package, the weights, GPU resources, and optionally an external MSA service all have to be available separately. Running Chai-1 alongside Boltz-2 buys a second model, not a second experiment. Both are all-atom diffusion co-folders of the same family, so a complex they both like is a complex they can be wrong about together, and the ipTM they agree on is still the models talking about themselves. What survives both is a shorter list to take to the bench. That is all it is.
A Skill is a directory of files, so installing one is copying that directory
somewhere an agent looks. With Node 18+ and git on PATH (npm resolves a
github: spec through it), and nothing cloned:
npx github:PKU-YuanGroup/OpenAI4S install chai1 --target claude
--target claude writes to ~/.claude/skills, claude-project to
./.claude/skills, openai4s to <data_dir>/user-skills, and --dir <path>
to anywhere you name. The resolved absolute path is printed before anything is
written there, and --dry-run stops at that plan. A reinstall refuses to
overwrite a copy you have edited or one it did not install, and uninstall
removes only the files it wrote. For a recipe included in the npm release,
npx @pku-yuangroup/openai4s-skills install chai1 --target claude installs the
published copy. The npm catalog can differ from this repository.
Without Node, take the directory itself — and turn it into a .zip if an upload
field wants one. The tarball is the whole repository (over 100 MB), and the pipe
is a POSIX shell recipe (macOS, Linux, WSL) that extracts only this directory:
curl -L https://codeload.github.com/PKU-YuanGroup/OpenAI4S/tar.gz/refs/heads/main \
| tar -xz --strip-components=2 OpenAI4S-main/skills/chai1
python3 -m zipfile -c chai1.zip chai1
The click-through form of that same download is the
repository zip,
which is also the route to take on Windows: extract it and copy skills/chai1/
out (unzip main.zip 'OpenAI4S-main/skills/chai1/*' unpacks only that
directory). If you already run OpenAI4S there is nothing to install — the wheel
ships every bundled Skill, and a bundled Skill takes precedence over a
same-named copy in <data_dir>/user-skills. Targets, provenance, and what the
installer refuses to do:
tools/skills-installer/.
| File | Responsibility |
|---|---|
SKILL.md |
Most of it is about one choice: MSA-backed run or the ESM-embedding path. Skipping the MSA server is faster but typically lands a few ipTM points behind, and it saves no GPU memory, because the embedding path loads a traced 3-billion-parameter ESM2 next to the trunk. Around that decision sit the >protein\|name=… header syntax for the multi-entity FASTA, the run_inference arguments, the ranked pred.model_idx_*.cif files and their scores.*.npz, and why an unset CHAI_DOWNLOADS_DIR either re-pulls 5 GB on every cold start or dies mid-run with a PermissionError. Ends on running Chai-1 against a second model for consensus, plus licensing. |