bio-inference — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited bio-inference (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
本范畴只处理“稳定执行已有模型推理”。如果用户要改模型层、训练、微调、datapipe adapter 或组件契约,读取 ../onescience-models/SKILL.md。
./af2-folding-inference/SKILL.md./af3-protenix-complex-inference/SKILL.md./simplefold-inference/SKILL.md./rfdiffusion-inference/SKILL.md./proteinmpnn-inference/SKILL.md./ptdit-protoken-inference/SKILL.md./evo2-inference/SKILL.md./molsculptor-inference/SKILL.mdreferences/model_inference_matrix.md:模型到入口、输入、checkpoint、输出和高危混淆点的矩阵。references/inference_execution_contract.md:所有生信模型推理必须遵守的 preflight、run、postflight 契约。../../../onescience-runtime/assets/bio_inference_templates/bio_inference_handoff.yaml:交给 onescience-skill -> onescience-coder/runtime 前的统一交接模板。{onescience_path}/onescience/examples/biosciences/_manifests/inference_run_manifest.yaml:单次推理运行的可复现 manifest。{onescience_path}/onescience/examples/biosciences/_manifests/model_requests/:OneScience 官方 per-model request 示例。{onescience_path}/onescience/examples/biosciences/_manifests/tools/validate_bio_inference_manifest.py:无第三方依赖地检查 manifest 是否缺关键字段。{onescience_path}/onescience/examples/biosciences/_manifests/tools/check_inference_outputs.py:按模型类型检查输出目录是否至少包含预期产物。model_family,再判定 inference_mode,不要从输出类型反推模型。例如“结构文件”可能来自 AlphaFold、OpenFold、AlphaFold3、Protenix、SimpleFold、RFdiffusion 或 ProToken。examples/biosciences/alphafold/run_alphafold.pyexamples/biosciences/openfold/run_pretrained_openfold.pyexamples/biosciences/alphafold3/run_alphafold.pyexamples/biosciences/protenix/runner/inference_unified.pyexamples/biosciences/simplefold/inference.py 或 sample.pyexamples/biosciences/RFdiffusion/scripts/run_inference.pyexamples/biosciences/ProteinMPNN/protein_mpnn_run.pyexamples/biosciences/pt_dit/example_scripts/de_novo_design.pyexamples/biosciences/evo2/infer.py 或 predict.pysrc/onescience/flax_models/MolSculptor/train/inference.py 或 examples/biosciences/molsculptor/inference/*onescience-skill 后让执行层进入 onescience-runtime 的 discover/preflight;若环境未 ready 再回退 onescience-installer,不要在角色层直接承诺可运行。examples/biosciences/_manifests/tools/,role 层只把工具路径写入交接摘要。bio_task_family: bio-inference
selected_concrete_skill:
model_family:
inference_mode:
input_protocol:
entrypoint:
checkpoint:
runtime_device:
preflight_checks:
command_plan:
expected_outputs:
output_validation_plan:
failure_recovery:
execution_entry:~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.