Hermes Benchmark Outer-Loop Harness Model Card
这是可索引项目报告证据页：它保留 Hermes Benchmark Outer-Loop Harness 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。
Hermes Benchmark Outer-Loop Harness Model Card
| Field | Value |
|---|---|
| Repository | howdymary/hermes-agent-metaharness |
| Category | Hermes Benchmark Outer-Loop Harness |
| Stars / forks snapshot | 89 / 9 |
| Commits / issues / PRs snapshot | 11 / 0 / 0 |
| Language | Python |
| License | MIT |
| Raw capture | raw-github/howdymary_hermes-agent-metaharness.md |
| Updated by | hourly public metadata update, 2026-06-04 21:38 +0800 |
1. Role in Self Evolve
Hermes Agent Meta-Harness is a standalone outer-loop evaluator that optimizes Hermes benchmark harness candidates without modifying model weights. It matters because self-evolving agents need explicit memory, harness, benchmark, and safety substrates before their improvement claims become trustworthy.
2. Working Principle
select candidate -> evaluate on TBLite/TB2 -> parse archives -> compare baseline vs candidate -> update frontier
3. Evidence Path
web-observed GitHub page showed 89 stars, 9 forks, 0 issues, 0 pull requests, 11 commits, MIT license, explicit TBLite/TB2 benchmark orchestration, baseline-vs-candidate comparison, and frontier-backed deterministic search. This iteration keeps freshness honest: the snapshot comes from the public GitHub page observed on 2026-06-04, while shell GitHub API access remained blocked in this workspace.
4. Teaching Use
Use this card to explain Hermes Benchmark Outer-Loop Harness: it shows how memory systems, harness maps, benchmark suites, and outer-loop evaluators connect to the broader self-evolving-agent pipeline.
5. Limits
The repository was not cloned in this iteration; no benchmark run, workflow execution, or agent loop experiment was executed. Counts and claims are visible public-page signals unless independently revalidated later.