Agent Zero Runtime Model Card
这是可索引项目报告证据页：它保留 Agent Zero Runtime 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 Zero Runtime Model Card
| Field | Value |
|---|---|
| Repository | agent0ai/agent-zero |
| Category | Autonomous Agent Runtime |
| Stars / forks snapshot | 17600 / 3600 |
| Language | Python |
| License | Public repository license visible |
| Raw capture | raw-github/agent0ai_agent-zero.md |
| Updated by | hourly public metadata update, 2026-05-26 04:41 +0800 |
1. Role in Self Evolve
Agent Zero is a production-facing runtime substrate: it combines Linux execution, memory, browser automation, tools, projects, skills, plugins, and subordinate agents, so it can host persistent improvement assets even when it is not itself a strict benchmarked self-evolution paper.
2. Working Principle
user goal -> project-scoped agent workspace -> tools / browser / memory / skills -> inspectable artifacts -> future tasks reuse the accumulated project state
3. Evidence Path
Web-observed GitHub evidence showed about 17.6k stars, 3.6k forks, 2,089 commits, v1.13 latest release on 2026-05-05, a full Linux environment, project-scoped skills, memory, browser automation, plugins, and multi-agent cooperation. Shell GitHub API freshness was not available in this run, so the metadata is web-observed rather than API-verified.
4. Teaching Use
Use this card to separate runtime capability from self-evolution proof: Agent Zero has many surfaces where memory and skills can improve, but claims of improvement still need independent evaluator evidence.
5. Limits
The repository was not cloned or executed locally. No security, sandbox, memory-quality, or multi-agent evaluation was run in this iteration.