Key Takeaways
- 107 projects in the OpenClaw science ecosystem trace back to 6 upstream codebases
- OpenClaw (TypeScript) is the largest parent: most projects fork or build on it directly
- NanoBot (Python) spawned the Python-native branch: MolClaw, lightweight research agents
- ZeroClaw (Rust) powers the performance-critical branch: DrugClaw (Rust), infrastructure tools
- PicoClaw (Go) enables the edge/IoT branch: field sensors, embedded devices
- Independent creations account for ~20% — built from scratch without forking any codebase
- Understanding lineage helps choose the right tool: fork-based projects inherit upstream features and bugs
What Is the Lineage Map?
The OpenClaw ecosystem has 107 projects, but they didn't all appear independently. Most share common ancestry — forked from or built upon a handful of upstream codebases. The lineage map traces these relationships.
Why it matters:
- Fork-based projects inherit upstream features (MCP, skills, multi-platform messaging)
- They also inherit upstream bugs and architectural constraints
- Independent projects may lack ecosystem integration but offer novel approaches
- Choosing a tool means understanding what it inherits
The 6 Upstream Codebases
| Codebase | Language | Stars | Descendants | Notable Forks |
|---|---|---|---|---|
| OpenClaw | TypeScript | 323K+ | ~60 projects | MedgeClaw, EdgeClaw, ScienceClaw variants |
| NanoBot | Python | 34K+ | ~15 projects | MolClaw, research-focused agents |
| ZeroClaw | Rust | 27K+ | ~8 projects | DrugClaw (Rust), infrastructure tools |
| PicoClaw | Go | 25K+ | ~5 projects | Edge/IoT deployments |
| NanoClaw | TypeScript | 24K+ | ~3 projects | Security-focused containers |
| MicroClaw | Rust | 576 | ~2 projects | Embedded sensor agents |
Lineage Branches
Branch 1: OpenClaw → Science Agents
The largest branch. Projects fork OpenClaw and add domain-specific skills:
- OpenClaw → MedgeClaw (biomedical, 140 K-Dense skills)
- OpenClaw → EdgeClaw (edge-cloud, privacy + cost routing)
- OpenClaw → ScienceClaw variants (4 independent projects, same name)
- OpenClaw → PaperClaw variants (6 independent projects)
- OpenClaw → OmicsClaw (multi-omics)
- OpenClaw → LabClaw (240 wet-lab skills)
Branch 2: NanoBot → Python Research
Python researchers who preferred NanoBot's lightweight approach:
- NanoBot → MolClaw (molecular science orchestrator)
- NanoBot → Various research-specific agents
Branch 3: Independent Creations
~20% of projects were built from scratch:
- EvoScientist (self-evolving multi-agent, Python)
- OriGene (target discovery, custom architecture)
- ChemCrow (LangChain-based, predates the ecosystem)
- Virtual Lab (Stanford, multi-agent protein design)
FAQ
Q1: How do I know if a project is a fork or independent?
Check the GitHub "forked from" indicator. If absent, check the README for dependency declarations. Our project pages on Claw4Science list the base framework for each tool.
Q2: Do forks automatically get upstream updates?
Not automatically. Most forks diverge over time. Some actively merge upstream changes (EdgeClaw tracks OpenClaw releases), others don't.
Q3: Should I prefer forks or independent projects?
Forks benefit from ecosystem integration (skills, messaging, MCP). Independent projects may offer novel approaches but lack compatibility. Choose based on your priority: ecosystem integration vs. unique capabilities.
Summary
The 107-project OpenClaw science ecosystem traces back to 6 upstream codebases, with OpenClaw (TypeScript) as the dominant parent. Understanding lineage helps researchers choose tools wisely — fork-based projects inherit ecosystem features, while independent creations offer novel approaches. The ecosystem is still young enough that lineage matters for predicting a project's trajectory.