EvoMap Evolver Model Card
这是可索引项目报告证据页：它保留 EvoMap Evolver 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。
EvoMap Evolver Model Card
One Sentence
EvoMap Evolver is a GEP-based self-evolution engine for AI agents, centered on auditable Genes, Capsules, and Events.
Three Sentences
It matters because it tries to make agent capability evolution inspectable rather than hidden in prompt edits. Its README surface includes CLI usage, multilingual docs, a SKILL.md integration path, and links to GEP documentation. In the project graph it belongs near skill systems, harness evolution, and memory/runtime projects.
Model Card
| Field | Value |
|---|---|
| Repository | EvoMap/evolver |
| Source | raw-github/evomap_evolver.md |
| Category | Auditable agent evolution engine |
| Pattern | Genome Evolution Protocol -> Genes/Capsules/Events -> logged evolution |
| Evidence | GitHub raw capture, 2026-05 |
Teaching Use
Use EvoMap to teach “evolution as a traceable protocol.” A reader can ask: what is the gene, what is the capsule, what event records the change, and how is the evolved behavior verified? This makes the project useful for bridging abstract self-evolution theory and operational logs.
Evidence And Limits
The raw capture reports about 7.5k stars, 760 forks, JavaScript/Node.js, GPL-3.0, documentation, and an npm package. The next step is deeper code analysis of src, examples, and SKILL.md to verify whether the GEP abstraction is enforced by code or mainly by documentation.