Crabbox Model Card
这是可索引项目报告证据页：它保留 Crabbox 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。
Crabbox Model Card
| Field | Value |
|---|---|
| Repository | openclaw/crabbox |
| Category | Browser Agent Benchmark and Evaluation Harness |
| Stars / forks snapshot | 539 / 61 |
| Language | Go |
| License | MIT |
| Raw capture | raw-github/openclaw_crabbox.md |
| Updated by | hourly public metadata update, 2026-05-30 01:15 +0800 |
1. Role in Self Evolve
crabbox is OpenClaw’s browser-agent benchmark framework for reproducible task execution and evaluator-driven comparison. It matters because self-evolving agents need repeatable harness control, measurable feedback loops, and reusable skill procedures before claiming stable improvement.
2. Working Principle
define browser-agent task suites -> execute agent runs with standardized tooling and policies -> score outcomes with reproducible evaluators -> provide benchmark signals for agent iteration and harness tuning
3. Evidence Path
web-observed GitHub page showed 539 stars, 61 forks, 945 commits, MIT license, and repository framing as a benchmark framework for browser agents. 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 Browser Agent Benchmark and Evaluation Harness: it shows how harness/runtime/benchmark layers convert agent behavior into reproducible and auditable engineering workflows.
5. Limits
The repository was not cloned in this iteration; no benchmark run, plugin install, workflow execution, or agent loop experiment was executed. Counts and claims are visible public-page/search signals unless independently revalidated later.