@@ -21,7 +21,7 @@
 
 ---
 
-ScienceEvolve is the fully autoresearch framework in paper [Rethinking Scientific Discovery in the Agentic Era](https://arxiv.org/abs/2607.03863). Build an Agent team, follow its execution, improve the task using real feedback, and authorize Main to evolve the workflow itself.
+ScienceEvolve is the fully autonomous research framework presented in [Rethinking Scientific Discovery in the Agentic Era](https://arxiv.org/abs/2607.03863). Build an Agent team, follow its execution, improve the task using real feedback, and authorize Main to evolve the workflow itself.
 
 <a id="sdk"></a>
 
@@ -71,7 +71,7 @@ The **task loop** runs a method, collects real feedback, improves the method, an
 
 ### Set Loop rounds and the exit Agent
 
-Main starts a finite-element mesh convergence experiment. Coder → Executor → Review runs for three complete rounds, then exits to the selected Writer Agent, which delivers the error analysis report (errors.md). Both the batch size and exit Agent are configurable. [Configure SDK Loops](sdk/docs/workflow-loops-zh.md) (Chinese).
+Main starts a finite-element mesh convergence experiment. Coder → Executor → Review runs for three complete rounds, then exits to the selected Writer Agent, which delivers the error analysis report (errors.md). Both the number of rounds and exit Agent are configurable. [Configure SDK Loops](sdk/docs/workflow-loops-zh.md) (Chinese).
 
 <div align="center">
 
 
@@ -1,7 +1,7 @@
 <h1 align="center">ScienceEvolve</h1>
 
 <p align="center">
-  <strong>在一个工作流里，组织、运行和持续改进多 Agent 科研任务。</strong>
+  <strong>构建、运行并持续改进自己的多 Agent 科研工作流，让工作流本身也能演化。</strong>
 </p>
 
 <p align="center">
@@ -262,7 +262,7 @@ try {
 <tr><td align="left"><a href="example/biomni-r0/workflow.json">Biomni-R0</a></td><td align="left"><a href="https://biomni.stanford.edu/blog/biomni-r0-technical-report" title="Biomni-R0 technical report">技术报告</a></td><td align="left"><a href="https://github.com/snap-stanford/Biomni">Biomni</a></td><td align="left"><a href="example/biomni-r0/workflow.json">JSON</a> · <a href="example/biomni-r0/requirement.txt">需求</a></td></tr>
 <tr><td align="left"><a href="example/chemical-coscientist/workflow.json">化学 Coscientist</a></td><td align="left"><a href="https://www.nature.com/articles/s41586-023-06792-0" title="Autonomous chemical research with large language models">Nature 2023</a></td><td align="left"><a href="https://github.com/gomesgroup/coscientist">coscientist</a></td><td align="left"><a href="example/chemical-coscientist/workflow.json">JSON</a> · <a href="example/chemical-coscientist/requirement.txt">需求</a></td></tr>
 <tr><td align="left"><a href="example/raven/workflow.json">Raven Host</a></td><td align="left"><a href="https://arxiv.org/abs/2609.33439" title="Raven: The Harness of Harnesses for Composable Agentic Intelligence">arXiv:2609.33439</a></td><td align="left"><a href="https://github.com/EverMind-AI/Raven">Raven</a></td><td align="left"><a href="example/raven/workflow.json">JSON</a> · <a href="example/raven/requirement.txt">需求</a></td></tr>
-<tr><td align="left"><a href="example/kosmos/workflow.json">Kosmos</a></td><td align="left"><a href="https://arxiv.org/abs/2511.02824">arXiv:2511.02824</a></td><td align="left"><a href="https://github.com/EdisonScientific/kosmos-figures">实验与分析资源</a></td><td align="left"><a href="example/kosmos/workflow.json">JSON</a> · <a href="example/kosmos/requirement.txt">需求</a></td></tr>
+<tr><td align="left"><a href="example/kosmos/workflow.json">Kosmos</a></td><td align="left"><a href="https://arxiv.org/abs/2511.02824">arXiv:2511.02824</a></td><td align="left"><a href="https://github.com/EdisonScientific/kosmos-figures">图表与分析资源</a></td><td align="left"><a href="example/kosmos/workflow.json">JSON</a> · <a href="example/kosmos/requirement.txt">需求</a></td></tr>
 <tr><td align="left"><a href="example/alphaevolve/workflow.json">AlphaEvolve</a></td><td align="left"><a href="https://arxiv.org/abs/2506.13131">arXiv:2506.13131</a></td><td align="left"><a href="https://github.com/google-deepmind/alphaevolve_repository_of_problems">问题与验证器</a></td><td align="left"><a href="example/alphaevolve/workflow.json">JSON</a> · <a href="example/alphaevolve/requirement.txt">需求</a></td></tr>
 </tbody></table>
 
@@ -298,6 +298,8 @@ SDK 打包使用 [`.agents/skills`](.agents/skills) 中的内置技能。对应
 
 内置版本包含 ScienceEvolve 的集成调整。项目角色技能 `research-director`、`idea-agent`、`experiment-runner`、`latex-writer` 和 `iclr-reviewer` 随本仓库维护。感谢上游作者与贡献者。
 
+CCFA 与 Nature 的内置副本未记录导入时的提交版本；上游仓库归属通过比对 Skill 名称与内容确认。
+
 <a id="configuration"></a>
 <a id="测试环境要求"></a>
 
 
@@ -34,7 +34,7 @@ scievolve workflow validate /absolute/workflow.json
 scievolve start /absolute/workflow.json /absolute/requirement.txt --data-dir /absolute/scienceevolve-data --workspace /absolute/task-project --gateway GATEWAY_ID
 ```
 
-The task directory must exist and be separate from `dataDir`. Git-off runs directly in it; Git-on uses isolated worktrees and delivers the integrated result back. Press Ctrl+C to pause, wait for shutdown, then use the printed resume command. Keep one SDK host per data directory.
+The task directory must exist, and both it and `dataDir` must use absolute paths. The two directories must be separate, and neither may contain the other. Git-off runs directly in the task directory; Git-on uses isolated worktrees and delivers the integrated result back. Press Ctrl+C to pause, wait for shutdown, then use the printed resume command. Keep one SDK host per data directory.
 
 ## Node API
 
 
@@ -34,7 +34,7 @@ scievolve workflow validate /absolute/workflow.json
 scievolve start /absolute/workflow.json /absolute/requirement.txt --data-dir /absolute/scienceevolve-data --workspace /absolute/task-project --gateway GATEWAY_ID
 ```
 
-任务目录须已存在，并与 `dataDir` 分开且互不包含。Git-off 在原目录执行；Git-on 使用隔离 worktree，集成后回写结果。按 Ctrl+C 暂停，等待退出，再使用打印出的命令恢复。同一个数据目录同时由一个 SDK 宿主持有。
+任务目录须已存在，且任务目录与 `dataDir` 均须使用绝对路径。两个目录分开且互不包含。Git-off 在任务目录中执行；Git-on 使用隔离 worktree，集成后回写结果。按 Ctrl+C 暂停，等待退出，再使用打印出的命令恢复。同一个数据目录同时由一个 SDK 宿主持有。
 
 ## Node API