CUGA Agent Model Card
这是可索引项目报告证据页：它保留 CUGA Agent 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。
CUGA Agent Model Card
| Field | Value |
|---|---|
| Repository | cuga-project/cuga-agent |
| Category | Enterprise Generalist Agent Harness |
| Stars / forks snapshot | 742 / 132 |
| Language | Python |
| License | Open source public repository |
| Raw capture | raw-github/cuga-project_cuga-agent.md |
| Updated by | hourly public metadata update, 2026-05-26 09:12 +0800 |
1. Role in Self Evolve
CUGA is an enterprise generalist agent harness that wires APIs, MCP servers, policies, memory, knowledge and agent skills into configurable web/API task execution. It matters because self-evolving agents need reliable harnesses, memory/skill surfaces, feedback evidence and benchmark loops before any improvement claim can be trusted.
2. Working Principle
enterprise agent config -> tools/MCP/OpenAPI -> policies/HITL -> optional memory/knowledge/skills -> trajectory visualization
3. Evidence Path
web-observed GitHub page showed 742 stars, 132 forks, 304 commits, Python/TypeScript stack, v0.2.28 latest release on 2026-05-10, README claims for MCP/OpenAPI/LangChain tools, policy/HITL, optional memory, knowledge and SKILL.md assets. Shell GitHub API access remained blocked by DNS and local gh auth was invalid, so this card treats the snapshot as web-observed rather than API-verified.
4. Teaching Use
Use this card to explain Enterprise Generalist Agent Harness: it shows which part of the agent improvement stack is made operational, and where evaluator, policy, memory, skill, task-state or artifact evidence enters the loop.
5. Limits
The repository was not cloned in this iteration; no benchmark, install flow, workflow run, policy audit, memory experiment, OpenClaw run, or agent loop was executed. Counts and claims are visible public-page/search signals unless independently revalidated later.