CodeScaleBench Model Card
这是可索引项目报告证据页：它保留 CodeScaleBench 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。
CodeScaleBench Model Card
| Field | Value |
|---|---|
| Repository | sourcegraph/CodeScaleBench |
| Category | Enterprise-Scale Coding Agent Benchmark Harness |
| Stars / forks snapshot | 25 / 3 |
| Language | Python |
| License | Apache-2.0 |
| Raw capture | raw-github/sourcegraph_codescalebench.md |
| Updated by | hourly public metadata update, 2026-05-29 04:05 +0800 |
1. Role in Self Evolve
CodeScaleBench is a benchmark suite for measuring coding agents with external retrieval tools on large enterprise-scale software tasks. It matters because self-evolving agents need repeatable harness control, measurable feedback loops, and reusable skill procedures before claiming stable improvement.
2. Working Principle
enterprise codebase tasks -> Harbor/Claude harness with baseline vs MCP retrieval configs -> dual-verifier scoring and cost tracking -> auditable snapshots for benchmark governance
3. Evidence Path
web-observed GitHub page showed 25 stars, 3 forks, 1,310 commits, Apache-2.0 license, and README describing 275 tasks with auditable run snapshots. 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-Scale Coding Agent Benchmark 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.