Add DeepRead evidence-first reading skill

[xiehuan123]

Summary

Adds DeepRead as an external Codex-compatible skill under Data & Analysis.

DeepRead handles a practical gap between generic summarization and source-grounded analysis: it requires important claims to be paired with evidence and source locations, marks unsupported statements explicitly, and can produce knowledge maps, cross-document comparisons, or Feynman review plans.

Codex usage

npx skills@latest add xiehuan123/dsh-deepread

Deep-read architecture.pdf in knowledge-map mode. Trace each important claim
to its evidence and page location, and mark anything the source does not support.

The skill has an English trigger description, no runtime dependencies, prompt-injection guidance for source documents, and real output examples in the project README. The same repository also provides an optional full DeepSeek Harness plugin, clearly documented as a separate runtime form.

[xiehuan123]

维护者好，补充一下 DeepRead 的最新状态：

• 已发布正式版 v1.0.0；
• 当前 29 Stars / 2 Forks；
• 已有 13 个相关社区 PR 合并；
• Codex 形态仍是可移植的 SKILL.md 工作流，不依赖 DeepSeek Harness 运行时；
• v1.0.0 补齐了统一 TypeScript 构建、跨宿主打包验证和更清晰的 Host/Skill 边界。

本 PR 当前显示可干净合并，尚无修改请求。由于上游已经升级到 1.0.0，请问是否希望我先同步最新技能快照，或维持当前提交供审查？我可以按目录规范继续调整。