MiniClaw Deep Dive Chapter 17: Relationship with OpenClaw
> 🔗 This chapter analyzes the relationship between MiniClaw and OpenClaw, including inherited core concepts and simplification differences.
---
17.1 Inherited Core Concepts
17.1.1 Microkernel Architecture
MiniClaw fully inherits OpenClaw's microkernel architecture philosophy, consisting of three layers:
- DNA layer (genes): templates. OpenClaw uses multiple template files; MiniClaw keeps 9 core templates.
- Kernel layer (brain): OpenClaw has a complex kernel; MiniClaw provides a streamlined kernel of roughly 2,400 lines.
- Interface layer (body): OpenClaw supports MCP plus multiple protocols; MiniClaw supports only MCP.
- 5 time modes:
morning(06:00–09:00),work(09:00–18:00),break(12:00–14:00),evening(18:00–22:00),night(22:00–06:00) — all also present in OpenClaw - Token budget management
- Priority compilation
- Session continuation detection
- Entry count > 20
- Log budget usage > 40%
- Oldest entry > 8 hours and entry count > 5
- Log size > 8KB
- Feature-complete, suited to production environments
- Highly customizable, with a learning cost
- Supports multiple protocols and deployment methods
- Minimal implementation lowering the entry barrier
- Retains core features, easy to understand
- One-command run, zero-config startup
- 🔗 Inherited concepts: microkernel architecture, ACE adaptive context engine, memory distillation, heartbeat and evolution
- 📊 Simplifications: 70%+ less code, core features only, MCP protocol only, one-command zero-config run
- 🎯 Positioning: OpenClaw = production-grade complete solution; MiniClaw = entry-level minimal solution
- 💡 Recommendations: beginners and learners of underlying principles → MiniClaw; production deployment → OpenClaw
17.1.2 ACE Engine
ACE (Adaptive Context Engine) follows the pipeline: time mode → context selection → token budget, with smart trimming, priority ordering, and dynamic compilation. MiniClaw inherits and simplifies the engine:
17.1.3 Memory Distillation
The distillation mechanism compresses short-term memory (daily logs of raw/temporary records) into long-term memory (MEMORY.md containing high-value facts and decision records), with a compression ratio of about 10:1.
Trigger conditions:
Distillation flow: 1. Analyze log content 2. Extract high-value information 3. Update MEMORY.md 4. Archive the original log
---
17.2 Simplification Differences
17.2.1 Feature Comparison
| Dimension | OpenClaw | MiniClaw |
|---|---|---|
| Code size | ~10,000+ lines | ~2,700 lines |
| Feature completeness | Full feature set | Core features |
| Deployment complexity | Environment configuration required | One-command npx run |
| Learning curve | Steep | Gentle |
| Customization | Highly customizable | Template-based configuration |
| Protocol support | MCP + others | MCP only |
| Use case | Production environments | Learning / quick trial |
17.2.2 Design Philosophy
OpenClaw:
MiniClaw:
Relationship: MiniClaw = a minimal implementation of OpenClaw's core concepts — effectively a "teaching" or "starter" edition of OpenClaw.
---
Chapter Summary
*This document is Chapter 17 of the "MiniClaw Deep Dive" series; the next chapter covers best practices in detail.*