Agent Smith Model Card
这是可索引项目报告证据页：它保留 Agent Smith Model Card 的源材料入口、机制线索和限制提醒；正文仍需 reader/editor 与 academic public-copy review 后才能当作最终结论引用。
Agent Smith Model Card
| Field | Value |
|---|---|
| Repository | cyijun/agent-smith |
| Category | Directory-Isolated Multi-Agent Protocol |
| Stars / forks snapshot | 18 / 2 |
| Language | Markdown/Skill Protocol |
| License | MIT |
| Raw capture | raw-github/cyijun_agent-smith.md |
| Updated by | hourly public metadata update, 2026-05-25 12:21 +0800 |
1. Role in Self Evolve
Agent Smith is a self-similar multi-agent collaboration protocol that decomposes work recursively while isolating each agent through inbox/private/outbox/children directories.
2. Working Principle
root task inbox -> recursive child agents -> directory isolation -> outbox aggregation
3. Evidence Path
web GitHub page observed 20 commits, MIT license, 18 stars and 2 forks; README describes recursive self-similar multi-agent execution, directory isolation, Claude Code skill entry, OpenCode/OpenClaw topics, max depth 3 and max 5 children constraints. Shell GitHub API access remained DNS-blocked and the local gh token was invalid in this run, so this card treats the current snapshot as web-observed rather than API-verified.
4. Teaching Use
Use this card to explain how Agent Smith fits the raw -> classification -> project card -> site/report pipeline. It is useful for comparing whether self-evolution is implemented as skill supply, memory substrate, planning persistence, harness governance, multi-agent orchestration, or benchmark/evaluation infrastructure.
5. Limits
当前未克隆源码，未运行 benchmark、skill install flows、memory experiments、MCP servers、agent evolution loops、security scanners 或 production deployments；star/fork/commit/release 快照来自公开 GitHub 页面文本或可见页面片段。