Repository navigation
What's New
- 🔧 Platform-agnostic refactor — all OpenCode-specific couplings removed (
task()/todowrite/background_output/run_in_background/[ALL BACKGROUND TASKS COMPLETE]) replaced with platform-agnostic dispatch. The same skill now runs on any AI coding tool: Claude Code, Codex CLI, Cursor, DSH, Windsurf, Cline, and more. - 🔧 Adaptive chapter writing — chapters are written in parallel by default when a multi-agent tool is detected; without one, the pipeline automatically falls back to sequential writing (same output, slightly slower). Research never blocks on parallelism.
- 📐 Neutralized platform-specific naming — Exa-specific manifest fields renamed to neutral equivalents (
exa_available→builtin_available,layer0_exa→layer0_builtin); "CLI built-in engine" wording unified to "tool built-in engine" across all docs. - 📐 Unified install guide — README / README_EN install section rewritten from a per-tool fork into a single universal guide plus a per-tool entry registration table.
- 🔧 De-coupled entry points —
command/research.mdno longer hardcodesskill(name=...);command/update.mddrop hardcoded~/.opencode/skillspaths;scrapling-mcp-server.pydocstring made platform-agnostic.
Verified end-to-end on DSH with a full quick-mode research run (outline → data collection → chapter writing → assembly → QA, all PASS).
🇨🇳 中文说明
- 🔧 平台无关重构 — 移除全部 OpenCode 专属耦合(
task()/todowrite/background_output/run_in_background/[ALL BACKGROUND TASKS COMPLETE]),改为平台无关派发。同一套 skill 现在可在任意 AI 编程工具上运行:Claude Code、Codex CLI、Cursor、DSH、Windsurf、Cline 等。 - 🔧 章节撰写自适应 — 探测到多 agent 工具时默认并行撰写;没有则自动降级为串行撰写(产物一致,仅耗时略增)。调研不因无法并行而中断。
- 📐 平台专属命名中性化 — Exa 相关 manifest 字段改为中性命名(
exa_available→builtin_available、layer0_exa→layer0_builtin);"CLI 内置引擎"表述统一为"工具内置引擎"。 - 📐 统一安装指引 — README/README_EN 安装章节从"按工具分叉"改写为"统一通用指引 + 各工具入口注册对照表"。
- 🔧 入口解耦 —
command/research.md不再硬编码skill(name=...);command/update.md去掉~/.opencode/skills硬路径;scrapling-mcp-server.py文档字符串改为平台无关。
已在 DSH 上以完整 quick 模式调研端到端验证(大纲 → 数据收集 → 章节撰写 → 装配 → QA 全部通过)。