AI Agent Architecture Analysis
Automated source-code-based architecture documentation for AI agent frameworks .
Generated by RepoSwarm — AI-powered multi-repo architecture discovery using Claude via Amazon Bedrock.
Agents Analyzed
| Agent | Source | Architecture |
|---|---|---|
| CoPaw | Python | CoPaw.arch.md |
| Hermes Agent | Python | hermes-agent.arch.md |
| IronClaw | Rust | ironclaw.arch.md |
| Letta | Python | letta.arch.md |
| MicroClaw | Various | microclaw.arch.md |
| Moltis | Go | moltis.arch.md |
| NanoClaw | TypeScript | nanoclaw.arch.md |
| Nanobot | Go | nanobot.arch.md |
| NemoClaw | Python | NemoClaw.arch.md |
| OpenClaw | TypeScript/Node.js | openclaw.arch.md |
| OpenFang | Rust | openfang.arch.md |
| PicoClaw | C/Rust | picoclaw.arch.md |
| ZeptoClaw | Zig | zeptoclaw.arch.md |
How This Was Generated
Each agent's source code was cloned and analyzed through 17 investigation steps covering:
- Project overview and entry points
- Architecture patterns and data flow
- Storage, state management, and persistence
- Security model and authentication
- API surface and protocols
- Deployment and configuration
- Dependencies and build system
- Testing and CI/CD
- And more...
Analysis performed by Claude Sonnet 4.6 via Amazon Bedrock on April 1–3, 2026.