AgentCures builds an integrated stack for planning, simulation, model inference, execution, evidence generation, and deployment in drug discovery programs.
flowchart TD
S[clawcures-ui] --> C[ClawCures]
C --> M[refua-mcp]
M --> R[refua]
R --> N[refua-notebook]
C --> D[refua-data]
C --> P[refua-preclinical]
C --> W[refua-wetlab]
C --> L[refua-clinical]
C --> H[refua-schema]
C --> G[refua-regulatory]
C --> B[refua-bench]
D --> T[tox21full]
S --> G
S --> L
S --> P
S --> D
X[refua-deploy] --> S
X --> C
X --> M
| Repository | Purpose | Key Features |
|---|---|---|
agentcures/ClawCures | Drug disovery agent (OpenClaw based) | OpenClaw planning, autonomous loops, policy checks, portfolio ranking |
agentcures/NanoCures | Drug discovery agent (NanoClaw based) | Container-isolated execution, per-group sessions, sender allowlists, translational workflow skills |
agentcures/refua | Core drug discovery ML toolkit | Boltz2 fold/affinity, BoltzGen design, unified Python API |
agentcures/refua-mcp | Typed MCP server for Refua tools | Structured tool contracts, fold/design/affinity, optional clinical and preclinical extras |
agentcures/clawcures-ui | Web control plane | Mission planning, job telemetry, trial and evidence workflows |
agentcures/refua-data | Data layer | Curated catalogs, API ingestion, local caching, parquet materialization |
agentcures/tox21full | Toxicity dataset generation | Full Tox21 multi-task dataset generation from public PubChem BioAssay records |
agentcures/refua-preclinical | Preclinical operations | GLP study planning, in vivo scheduling, bioanalysis pipelines |
agentcures/refua-wetlab | Wet lab operations | Experiment orchestration, assay tracking, sample lifecycle coordination |
agentcures/refua-clinical | Clinical simulation | PK/PD virtual patients, adaptive trial simulation, protocol optimization |
agentcures/refua-schema | Portfolio object model | Portfolio -> Disease -> Rationale -> Drug hierarchy, reused Refua objects, JSON/YAML round-tripping |
agentcures/refua-regulatory | Regulatory evidence and audit | Decision lineage, signed bundles, checklist gates |
agentcures/refua-bench | Benchmarking and regression gates | Statistical comparisons, provenance capture, baseline promotion |
agentcures/refua-notebook | Notebook visualization layer | Rich IPython and Jupyter rendering for Refua objects |
agentcures/refua-deploy | Deployment generation | Kubernetes/Compose/single-machine bundles for cloud and on-prem |
- Define mission intent in
clawcures-uiorClawCures. - Build candidate evidence and portfolio objects through
refua,refua-mcp,refua-data,tox21full, andrefua-schema. - Simulate wet lab, preclinical, and clinical strategy with
refua-wetlab,refua-preclinical, andrefua-clinical. - Gate quality with
refua-bench. - Package traceable evidence with
refua-regulatory. - Render and apply runtime bundles with
refua-deploy.
- Website: https://www.agentcures.com
- News: https://www.agentcures.com/news