SICA Self-Improving Coding Agent Model Card
这是可索引项目报告证据页：它保留 SICA Self-Improving Coding 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。
SICA Self-Improving Coding Agent Model Card
| Field | Value |
|---|---|
| Repository | MaximeRobeyns/self_improving_coding_agent |
| Category | Self-Improving Coding Agent |
| Stars / forks snapshot | 324 / 56 |
| Language | Python |
| License | MIT |
| Raw capture | raw-github/maximerobeyns_self_improving_coding_agent.md |
| Updated by | hourly public metadata update, 2026-05-26 04:41 +0800 |
1. Role in Self Evolve
SICA is a core self-evolution artifact because the target of improvement is the coding agent’s own codebase, not only a task answer produced by a fixed agent.
2. Working Principle
coding task -> agent modifies candidate code -> tests / review signal -> accepted patch -> agent codebase becomes the next starting point
3. Evidence Path
Web-observed GitHub evidence showed 324 stars, 56 forks, MIT license, Python-dominant implementation, and a citation to “SICA: A Self-Improving Coding Agent” from the ICLR 2025 Workshop on Scaling Self-Improving Foundation Models. 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 when explaining the difference between ordinary coding agents and self-referential coding agents: SICA changes the system that will perform future coding work.
5. Limits
The repository was not cloned and no SWE-Bench or local tests were run in this iteration. The model card only promotes public metadata and the improvement-loop role into the site.