Skip to content
dsh-hot
Release

fix(benchmark): 补齐S类诊断题的LLM评分并细化rubric

Source excerpt

改了什么 S13、S14、S20原来仍按关键词打分,这次接入现有的LLM判分器。S11、S16、S17、S18、S19、S21、S22沿用#212已合入的迁移,进一步写清每项怎么给分、什么情况给半分、哪些错误建议会触发封顶。10题版本统一升到4.1.0,题面和fixture保持原样。

Automatically summarized from source discussions. Sources and the full record are available below.

Review record and full evidence

Evidence & timeline

Verification

Not yet verified

Maintainer response

Untriaged

Author response

No response recorded

Earliest source found so far

Unknown

Sep 12, 2026, 03:32 PM

Research & review record

Lifecycle

Emerging

Source scope

16 automatically linked original sources; source relationships and independence are not human reviewed

Attribution confidence · 0%

First found by this site · Sep 12, 2026, 03:58 PM

Evidence, test conditions and updates
Research & review record · Reviewed propagation graph
06

Reviewed propagation graph

00

No source-to-source propagation relation recorded.

Source timeline

  1. Blog

    What Is DeepSeek Harness? The Agent Where Everything Is a Plugin | DataCamp

    Open source Opens in a new tab
  2. Blog

    DSH plugin boot race: non-atomic cordis.yml rewrite

    Open source Opens in a new tab
  3. GitHub PR

    fix(benchmark): 补齐S类诊断题的LLM评分并细化rubric

    Open source Opens in a new tab
  4. GitHub PR

    Comment on fix(benchmark): 补齐S类诊断题的LLM评分并细化rubric

    Open source Opens in a new tab
  5. GitHub PR

    docs(troubleshooting): a bundle-owned tool row leaks into every preset

    Open source Opens in a new tab
  6. GitHub PR

    feat(benchmark): add S24 dock-cotenant-crash trap (shared error boundary co-tenant unmount)

    Open source Opens in a new tab
  7. GitHub PR

    docs(skill): plugin-runtime-debug browser forensics reference + engine-divergence symptom family

    Open source Opens in a new tab
  8. GitHub PR

    Comment on docs(troubleshooting): scope the tool-registration-plane fix by version

    Open source Opens in a new tab
  9. GitHub PR

    Comment on fix(references): retire pre-0.1.7 package coordinates from the probe docs

    Open source Opens in a new tab
  10. GitHub PR

    fix: initialize and verify native runtime profiles

    Open source Opens in a new tab
  11. GitHub PR

    fix(smoke): fail release checks on unsuccessful shutdown

    Open source Opens in a new tab
  12. GitHub PR

    fix: stop H8 command process groups on timeout

    Open source Opens in a new tab
  13. GitHub PR

    Comment on fix(smoke): fail release checks on unsuccessful shutdown

    Open source Opens in a new tab
  14. GitHub PR

    Comment on fix: initialize and verify native runtime profiles

    Open source Opens in a new tab
  15. GitHub PR

    Comment on fix: initialize and verify native runtime profiles

    Open source Opens in a new tab
  16. GitHub PR

    fix(readme): correct stale root README counts, surface jump cards, and gate the drift

    Open source Opens in a new tab

Related stories

  • BenchmarkNot yet verified

    Investigation lead: fix(benchmark): fetch Web probe HTML from the discovered host origin

    The lead source, “fix(benchmark): fetch Web probe HTML from the discovered host origin,” says: “M5 and H8 exchange the bootstrap token from the URL printed by DSH, then fetch HTML from a fixed port 3080. If DSH prints a URL on another port, the probe reads a different endpoin.” 35 related sources are linked for comparison to determine whether they describe the same event.

    41 sources· GitHub / Linux DO / Blog
  • BenchmarkNot yet verified

    Investigation lead: benchmark: step5-preview S1-S24 双臂运行归档 + 双判分结果

    The lead source, “benchmark: step5-preview S1-S24 双臂运行归档 + 双判分结果,” says: “StepFun Step 5 Preview(OpenCode Zen 供应,公开发布名;预览端点背后精确 checkpoint 未经独立验证;opencode 编排)在 S 系列基准上的完整 24 题双臂运行、双判分与重跑 follow-up 归档。.” This is currently the only captured source.

    4 sources· GitHub
  • BenchmarkNot yet verified

    Investigation lead: feat(benchmark): S18 terminal sprite render trap (half-block SGR leak, trim ghosting, timer pinning)

    The lead source, “feat(benchmark): S18 terminal sprite render trap (half-block SGR leak, trim ghosting, timer pinning),” says: “Summary Adds benchmark task S18-terminal-sprite-render-trap (static, read-only report): a terminal half-block sprite renderer whose three independent rendering bugs (SGR background.” 36 related sources are linked for comparison to determine whether they describe the same event.

    42 sources· GitHub / Blog