scientify 1.11.0 → 1.12.1

package/skills/idea-generation/SKILL.md

@@ -19,32 +19,27 @@ Generate innovative research ideas grounded in literature analysis. This skill r
 
 **Core principle:** Ideas MUST be grounded in actual papers, not generated from model knowledge.
 
-**Workspace:** See `../_shared/workspace-spec.md` for directory structure. Outputs go to `$WORKSPACE/ideas/`.
+**Workspace:** `$W` = working directory provided in task parameter. Outputs go to `$W/ideas/`.
 
 ---
 
 ## Step 1: Check Workspace Resources
 
-First, check what resources already exist:
+First, check what resources already exist in `$W`:
 
 ```bash
-# Check active project
-cat ~/.openclaw/workspace/projects/.active 2>/dev/null
-
-# Check papers
-ls ~/.openclaw/workspace/projects/*/papers/ 2>/dev/null | head -20
-
-# Check survey results
-cat ~/.openclaw/workspace/projects/*/survey/clusters.json 2>/dev/null | head -5
+ls $W/papers/ 2>/dev/null | head -20
+ls $W/papers/_meta/ 2>/dev/null | head -10
+ls $W/survey/ 2>/dev/null
 ```
 
 ### Assess Available Resources
 
 | Resource | Location | Status |
 |----------|----------|--------|
-| Papers | `$WORKSPACE/papers/` | Count: ? |
-| Survey clusters | `$WORKSPACE/survey/clusters.json` | Exists: Y/N |
-| Repos | `$WORKSPACE/repos/` | Count: ? |
+| Papers | `$W/papers/` | Count: ? |
+| Survey clusters | `$W/survey/clusters.json` | Exists: Y/N |
+| Repos | `$W/repos/` | Count: ? |
 
 ---
 
@@ -57,14 +52,14 @@ Based on workspace state, ask user:
 >
 > Options:
 > 1. **Use existing papers** - Generate ideas from current collection
-> 2. **Search more** - Run `/literature-survey` to expand collection
+> 2. **Search more** - Run `/research-collect` to expand collection
 > 3. **Quick search** - Add 5-10 more papers on specific topic
 
 **If no papers:**
 > 📭 No papers found in workspace.
 >
 > To generate grounded ideas, I need literature. Options:
-> 1. **Run /literature-survey** - Comprehensive search (100+ papers, recommended)
+> 1. **Run /research-collect** - Comprehensive search (100+ papers, recommended)
 > 2. **Quick search** - Fetch 10-15 papers on your topic now
 > 3. **You provide papers** - Point me to existing PDFs/tex files
 
@@ -72,17 +67,17 @@ Based on workspace state, ask user:
 
 ## Step 3: Acquire Resources (if needed)
 
-### Option A: Delegate to /literature-survey (Recommended)
+### Option A: Delegate to /research-collect (Recommended)
 
 If user wants comprehensive search:
 ```
-Please run: /literature-survey {topic}
+Please run: /research-collect {topic}
 
 This will:
 - Search 100+ papers systematically
 - Filter by relevance (score ≥4)
 - Cluster into research directions
-- Save to $WORKSPACE/papers/
+- Save to $W/papers/
 
 After survey completes, run /idea-generation again.
 ```
@@ -101,21 +96,21 @@ Arguments:
 
 2. **Clone 3-5 reference repos:**
 ```bash
-mkdir -p $WORKSPACE/repos
-git clone --depth 1 {repo_url} $WORKSPACE/repos/{name}
+mkdir -p $W/repos
+git clone --depth 1 {repo_url} $W/repos/{name}
 ```
 
 3. **Download paper sources:**
 ```bash
-mkdir -p $WORKSPACE/papers/{arxiv_id}
-curl -L "https://arxiv.org/src/{arxiv_id}" | tar -xz -C $WORKSPACE/papers/{arxiv_id}
+mkdir -p $W/papers/{arxiv_id}
+curl -L "https://arxiv.org/src/{arxiv_id}" | tar -xz -C $W/papers/{arxiv_id}
 ```
 
 ---
 
 ## Step 4: Analyze Literature
 
-**Prerequisites:** At least 5 papers in `$WORKSPACE/papers/`
+**Prerequisites:** At least 5 papers in `$W/papers/`
 
 ### 4.1 Read Papers
 
@@ -135,7 +130,7 @@ Look for:
 - Scalability issues
 - Assumptions that could be relaxed
 
-Document gaps in `$WORKSPACE/ideas/gaps.md`:
+Document gaps in `$W/ideas/gaps.md`:
 ```markdown
 # Research Gaps Identified
 
@@ -151,7 +146,7 @@ Document gaps in `$WORKSPACE/ideas/gaps.md`:
 
 ## Step 5: Generate 5 Ideas
 
-Create `$WORKSPACE/ideas/idea_1.md` through `idea_5.md` using template in `references/idea-template.md`.
+Create `$W/ideas/idea_1.md` through `idea_5.md` using template in `references/idea-template.md`.
 
 **Requirements:**
 - Each idea cites ≥2 papers by arXiv ID
@@ -180,7 +175,7 @@ Create `$WORKSPACE/ideas/idea_1.md` through `idea_5.md` using template in `refer
 
 ### 6.2 Enhance Selected Idea
 
-Create `$WORKSPACE/ideas/selected_idea.md` with:
+Create `$W/ideas/selected_idea.md` with:
 - Detailed math (loss functions, gradients)
 - Architecture choices
 - Hyperparameters
@@ -210,13 +205,13 @@ Map idea concepts to reference implementations.
 
 See `references/code-mapping.md` for template.
 
-**Output:** `$WORKSPACE/ideas/implementation_report.md`
+**Output:** `$W/ideas/implementation_report.md`
 
 ---
 
 ## Step 8: Summary
 
-Create `$WORKSPACE/ideas/summary.md`:
+Create `$W/ideas/summary.md`:
 - All 5 ideas with scores
 - Selected idea details
 - Next steps: `/research-pipeline` to implement
@@ -236,6 +231,6 @@ Create `$WORKSPACE/ideas/summary.md`:
 
 ## Integration
 
-- **Before:** `/literature-survey` to collect papers
+- **Before:** `/research-collect` to collect papers
 - **After:** `/research-pipeline` to implement selected idea
 - **Alternative:** `/write-review-paper` to write survey instead

package/skills/metabolism/SKILL.md

@@ -0,0 +1,118 @@
+---
+name: metabolism
+description: "Knowledge metabolism cycle: ingest new papers, update knowledge state, detect cross-topic links, generate hypotheses. Use /metabolism to trigger manually."
+user-invokable: true
+---
+
+# Continuous Knowledge Metabolism — Incremental Cycle
+
+你正在执行知识新陈代谢循环。严格按以下步骤执行。
+
+**前提：** `metabolism/config.json` 必须已存在且 `currentDay >= 1`。如果不存在或 `currentDay` 为 0，提示用户先执行 /metabolism-init 完成初始化。
+
+## 准备
+
+1. 读取 `metabolism/config.json` 获取关键词、arXiv 分类、`processed_ids` 和 `currentDay`
+2. 读取 `metabolism/knowledge/_index.md` 获取当前知识状态
+
+## Step 1: Search（增量搜索）
+
+用滑动窗口（过去 5 天）搜索，靠 `processed_ids` 去重：
+
+```
+arxiv_search({
+  query: "{keywords} AND cat:{category}",
+  date_from: "{5天前 YYYY-MM-DD}",
+  sort_by: "submittedDate",
+  max_results: 30
+})
+
+openalex_search({
+  query: "{keywords}",
+  filter: "from_publication_date:{5天前 YYYY-MM-DD}",
+  sort: "publication_date",
+  max_results: 20
+})
+```
+
+合并结果，按 arXiv ID / DOI 去重，**跳过 `processed_ids` 中已有的论文**。
+
+下载新论文：
+
+```
+arxiv_download({ arxiv_ids: ["{id1}", "{id2}", ...] })
+unpaywall_download({ dois: ["{doi1}", "{doi2}", ...] })
+```
+
+## Step 2: Read（阅读）
+
+对每篇新论文：
+- 读 .tex 源码（优先）或 PDF
+- 提取：核心方法、关键结论、与已有知识的关系
+
+将每篇论文的 arXiv ID / DOI 追加到 `config.json` 的 `processed_ids`。
+
+## Step 3: Update Knowledge
+
+读取当前 `metabolism/knowledge/_index.md` 和相关 `topic-*.md`，根据今日阅读的论文更新。
+
+**更新原则：**
+- 新发现 → 添加到相关章节
+- 印证已有认知 → 补充证据来源
+- 与已有认知矛盾 → 标注分歧，保留两方论据
+- 跨领域关联 → 记录连接
+
+**篇幅管理：** 每个 topic 文件控制在 200 行以内。接近上限时，压缩早期内容（合并相似结论、删除低价值条目），保留信息密度。不要为了压缩而丢失关键结论和来源引用。
+
+## Step 4: Hypothesize（假设）
+
+更新完 knowledge.md 后，回顾今日新增内容，自问：
+
+- 有没有反复出现但尚未被验证的模式？
+- 有没有两个独立发现组合后暗示的新方向？
+- 有没有现有方法的明显空白？
+
+**有想法** → 写入 `metabolism/hypotheses/hyp-{NNN}.md`：
+
+```markdown
+# Hypothesis {NNN}
+
+## 假设
+{一句话}
+
+## 推理过程
+{基于哪些论文/知识得出，2-3 段}
+
+## 来源论文
+- {arxiv_id}: {title}
+
+## 自评
+- 新颖性: {1-5}
+- 可行性: {1-5}
+- 影响力: {1-5}
+```
+
+然后用 `sessions_send` 通知 main session。
+
+**没有想法** → 跳过，不要硬凑。
+
+## Step 5: Log & Finish
+
+写入 `metabolism/log/{YYYY-MM-DD}.md`：
+
+```markdown
+# Day {currentDay} — {YYYY-MM-DD}
+
+新论文: {N} 篇
+知识更新: {简述主要变更}
+假设: {有/无}
+```
+
+更新 `config.json`：`currentDay` +1。
+
+## 行为约束
+
+1. 不捏造论文中未出现的事实性声明，但可以用自身知识做推理和关联判断
+2. 没有想法时不生成假设
+3. 自主运行，不向人类提问
+4. 修改知识文件前必须先读取当前内容

package/skills/metabolism-init/SKILL.md

@@ -0,0 +1,80 @@
+---
+name: metabolism-init
+description: "Initialize knowledge metabolism for a research topic: broad literature survey, build baseline knowledge state, set up metabolism workspace"
+user-invokable: true
+---
+
+# Metabolism Initialization — Day 0 Baseline Building
+
+你正在为一个研究方向执行知识新陈代谢的初始化。这是 Day 0：构建领域基线知识。
+
+## 准备
+
+1. 检查 `metabolism/config.json` 是否存在
+   - 如果不存在：询问用户研究方向，创建 `metabolism/config.json`（包含 `keywords`, `categories`, `currentDay: 0`, `processed_ids: []`）
+   - 如果已存在且 `currentDay > 0`：提示用户已完成初始化，无需重复执行
+2. 创建目录结构（如不存在）：
+   ```
+   metabolism/
+     knowledge/
+     hypotheses/
+     experiments/
+     conversations/
+     log/
+   ```
+
+## Step 1: 宽泛调研
+
+委托 /research-collect 执行宽泛调研（不限日期），构建初始知识：
+
+```
+sessions_spawn({
+  task: "/research-collect\n研究主题: {从 config.json 的 keywords 提取}\n这是 Day 0 基线构建，请做宽泛调研（不限日期），覆盖领域经典工作和近期进展。\n预期产出: papers/_meta/*.json + papers/_downloads/",
+  label: "Day 0 Baseline Survey",
+  runTimeoutSeconds: 1800
+})
+```
+
+spawned session 共享工作目录，无需传路径。等待完成后，读取 `papers/_meta/*.json` 获取论文列表。
+
+## Step 2: 阅读与知识提取
+
+对每篇论文：
+- 读 .tex 源码（优先）或 PDF
+- 提取：核心方法、关键结论、领域现状
+
+将每篇论文的 arXiv ID / DOI 追加到 `metabolism/config.json` 的 `processed_ids`。
+
+## Step 3: 构建初始知识状态
+
+创建 `metabolism/knowledge/_index.md`：
+- Research Goal（从 config.json 提取）
+- Topics 表格（按论文主题聚类）
+- Cross-topic Links（如有）
+- Timeline（记录 Day 0）
+
+为每个识别出的主题创建 `metabolism/knowledge/topic-{name}.md`：
+- 已知方法
+- 关键论文与结论
+- 开放问题
+
+## Step 4: 记录日志
+
+写入 `metabolism/log/{YYYY-MM-DD}-init.md`：
+
+```markdown
+# Day 0 — Initialization
+
+日期: {YYYY-MM-DD}
+论文: {N} 篇
+主题: {列出识别的主题}
+状态: 基线构建完成
+```
+
+更新 `metabolism/config.json`：`currentDay` 设为 1。
+
+## 行为约束
+
+1. 不捏造论文中未出现的事实性声明
+2. 自主运行，不向人类提问（除初始配置外）
+3. 修改知识文件前必须先读取当前内容（如存在）

package/skills/{literature-survey → research-collect}/SKILL.md

@@ -1,6 +1,6 @@
 ---
-name: literature-survey
-description: "Use this when the user wants to find, download, or collect academic papers on a topic. Searches arXiv, filters by relevance, downloads PDFs and sources, clusters by research direction."
+name: research-collect
+description: "[Read when prompt contains /research-collect]"
 metadata:
   {
     "openclaw":
@@ -14,10 +14,12 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
+**Workspace:** `$W` = working directory provided in task parameter.
+
 ## Output Structure
 
 ```
-~/.openclaw/workspace/projects/{project-id}/
+$W/
 ├── survey/
 │   ├── search_terms.json      # 检索词列表
 │   └── report.md              # 最终报告
@@ -38,17 +40,13 @@ metadata:
 
 ### Phase 1: 准备
 
+确保工作目录结构存在：
+
 ```bash
-ACTIVE=$(cat ~/.openclaw/workspace/projects/.active 2>/dev/null)
-if [ -z "$ACTIVE" ]; then
-  PROJECT_ID="<topic-slug>"
-  mkdir -p ~/.openclaw/workspace/projects/$PROJECT_ID/{survey,papers/_downloads,papers/_meta}
-  echo "$PROJECT_ID" > ~/.openclaw/workspace/projects/.active
-fi
-PROJECT_DIR="$HOME/.openclaw/workspace/projects/$(cat ~/.openclaw/workspace/projects/.active)"
+mkdir -p "$W/survey" "$W/papers/_downloads" "$W/papers/_meta"
 ```
 
-生成 4-8 个检索词，保存到 `survey/search_terms.json`。
+生成 4-8 个检索词，保存到 `$W/survey/search_terms.json`。
 
 ---
 
@@ -76,13 +74,13 @@ arxiv_search({ query: "<term>", max_results: 30 })
 ```
 arxiv_download({
   arxiv_ids: ["<有用的论文ID>"],
-  output_dir: "$PROJECT_DIR/papers/_downloads"
+  output_dir: "papers/_downloads"
 })
 ```
 
 #### 2.4 写入元数据
 
-为每篇下载的论文创建元数据文件 `papers/_meta/{arxiv_id}.json`：
+为每篇下载的论文创建元数据文件 `$W/papers/_meta/{arxiv_id}.json`：
 
 ```json
 {
@@ -105,7 +103,7 @@ arxiv_download({
 
 #### 3.1 选择高分论文
 
-读取 `papers/_meta/` 下得分 ≥4 的论文，选出 **Top 5** 最相关论文。
+读取 `$W/papers/_meta/` 下得分 ≥4 的论文，选出 **Top 5** 最相关论文。
 
 #### 3.2 搜索参考仓库
 
@@ -116,41 +114,32 @@ arxiv_download({
 
 使用 `github_search` 工具：
 ```javascript
-// 示例：
 github_search({
   query: "{paper_title} implementation",
   max_results: 10,
   sort: "stars",
-  language: "python"  // 可选：根据论文领域选择语言
-})
-
-// 如果有具体方法名：
-github_search({
-  query: "{method_name} {author_last_name}",
-  max_results: 5
+  language: "python"
 })
 ```
 
-**提示**：如果需要 GitHub API 高频率限制，设置环境变量 `GITHUB_TOKEN`。
-
 #### 3.3 筛选与 clone
 
 对搜索到的仓库，评估：
 - Star 数（建议 >100）
 - 代码质量（有 README、有 requirements.txt、代码结构清晰）
-- 与论文的匹配度（README 中引用了论文 / 实现了论文中的方法）
+- 与论文的匹配度
 
-选择 **3-5 个**最相关的仓库，clone 到 `repos/`：
+选择 **3-5 个**最相关的仓库，clone 到 `$W/repos/`：
 
 ```bash
-mkdir -p "$PROJECT_DIR/repos"
-cd "$PROJECT_DIR/repos"
+mkdir -p "$W/repos"
+cd "$W/repos"
 git clone --depth 1 <repo_url>
 ```
 
 #### 3.4 写入选择报告
 
-创建 `$PROJECT_DIR/prepare_res.md`：
+创建 `$W/prepare_res.md`：
 
 ```markdown
 # 参考仓库选择
@@ -179,7 +168,7 @@ git clone --depth 1 <repo_url>
 #### 4.1 读取所有元数据
 
 ```bash
-ls $PROJECT_DIR/papers/_meta/
+ls $W/papers/_meta/
 ```
 
 读取所有 `.json` 文件，汇总论文列表。
@@ -191,15 +180,15 @@ ls $PROJECT_DIR/papers/_meta/
 #### 4.3 创建文件夹并移动
 
 ```bash
-mkdir -p "$PROJECT_DIR/papers/data-driven"
-mv "$PROJECT_DIR/papers/_downloads/2401.12345" "$PROJECT_DIR/papers/data-driven/"
+mkdir -p "$W/papers/data-driven"
+mv "$W/papers/_downloads/2401.12345" "$W/papers/data-driven/"
 ```
 
 ---
 
 ### Phase 5: 生成报告
 
-创建 `survey/report.md`：
+创建 `$W/survey/report.md`：
 - 调研概要（检索词数、论文数、方向数）
 - 各研究方向概述
 - Top 10 论文
@@ -222,3 +211,4 @@ mv "$PROJECT_DIR/papers/_downloads/2401.12345" "$PROJECT_DIR/papers/data-driven/
 |------|---------|
 | `arxiv_search` | 搜索论文（无副作用） |
 | `arxiv_download` | 下载 .tex/.pdf（需绝对路径） |
+| `github_search` | 搜索参考仓库 |

package/skills/research-experiment/SKILL.md

@@ -15,7 +15,7 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
-**Workspace:** See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+**Workspace:** `$W` = working directory provided in task parameter.
 
 ## Prerequisites
 

package/skills/research-implement/SKILL.md

@@ -15,7 +15,7 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
-**Workspace:** See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+**Workspace:** `$W` = working directory provided in task parameter.
 
 ## Prerequisites
 

package/skills/research-pipeline/SKILL.md

@@ -94,22 +94,17 @@ task 必须以 `/skill-name` 开头（触发 slash command 解析），后续行
 
 ## Workspace
 
-See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+`$W` = agent workspace root (see AGENTS.md for layout).
 
 ---
 
 ## Step 0: 初始化
 
-```bash
-ACTIVE=$(cat ~/.openclaw/workspace/projects/.active 2>/dev/null)
-```
+`$W` 即当前 agent 的工作目录（AGENTS.md 中定义）。
 
-如果没有 active project：
-1. 问用户：研究主题是什么？
-2. 创建项目目录
-3. 写入 `task.json`
+检查 `$W/SOUL.md` 是否包含研究方向信息。如果没有（BOOTSTRAP 未完成），提示用户先完成 BOOTSTRAP 配置。
 
-设置 `$W = ~/.openclaw/workspace/projects/{project-id}`
+确保 `$W` 下存在必要的子目录（如 `survey/`, `papers/` 等）。
 
 ---
 
@@ -122,8 +117,8 @@ ACTIVE=$(cat ~/.openclaw/workspace/projects/.active 2>/dev/null)
 **检查:** `$W/papers/_meta/` 目录存在且有 `.json` 文件？
 
 **如果缺失，调用 sessions_spawn 工具（然后停止，等待完成通知）：**
-- task: `"/literature-survey\n工作目录: {$W绝对路径}\n研究主题: {从task.json提取}\n请搜索、筛选、下载论文到工作目录的 papers/ 下。"`
-- label: `"Literature Survey"`
+- task: `"/research-collect\n工作目录: {$W绝对路径}\n研究主题: {从task.json提取}\n请搜索、筛选、下载论文到工作目录的 papers/ 下。"`
+- label: `"Research Collect"`
 - runTimeoutSeconds: `1800`
 
 **验证:** `ls $W/papers/_meta/*.json` 至少有 3 个文件

package/skills/research-plan/SKILL.md

@@ -14,7 +14,7 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
-**Workspace:** See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+**Workspace:** `$W` = working directory provided in task parameter.
 
 ## Prerequisites
 
@@ -23,8 +23,8 @@ metadata:
 | `$W/task.json` | /research-pipeline or user |
 | `$W/survey_res.md` | /research-survey |
 | `$W/notes/paper_*.md` | /research-survey |
-| `$W/repos/` | /literature-survey Phase 3 |
-| `$W/prepare_res.md` | /literature-survey Phase 3 |
+| `$W/repos/` | /research-collect Phase 3 |
+| `$W/prepare_res.md` | /research-collect Phase 3 |
 
 **If `survey_res.md` is missing, STOP:** "需要先运行 /research-survey 完成深度分析"
 

package/skills/research-review/SKILL.md

@@ -15,7 +15,7 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
-**Workspace:** See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+**Workspace:** `$W` = working directory provided in task parameter.
 
 ## Prerequisites
 

package/skills/research-subscription/SKILL.md

@@ -25,6 +25,13 @@ Use this skill when the user asks for:
 
 Do not stop at explanation.
 Create a real cron job via `scientify_cron_job`.
+Do not claim "still running/in progress" unless you created a real async handle (cron job id or task id). If no handle exists, finish the run in the same turn.
+For high-effort strict-quality runs that are unlikely to finish in one turn, start real async execution instead of replying "not executed":
+- if user forbids schedule creation, prefer `sessions_spawn` and return task id
+- otherwise prefer `scientify_cron_job` with `run_now: true` and return job id
+If current turn is already cron-triggered, never call `scientify_cron_job` again from inside that run (avoid nested cron/run_now recursion).
+When using `scientify_literature_state`, keep `scope/topic` consistent across prepare -> record -> status (reuse prepare output; do not replace scope with project id).
+If user gave hard constraints (for example exact/min core paper count), do not return status `ok` unless satisfied; otherwise persist `degraded_quality` with unmet reasons.
 
 ## Tool to call
 
@@ -33,6 +40,7 @@ Create a real cron job via `scientify_cron_job`.
 - `action: "upsert"`: create or update a schedule
 - `action: "list"`: show current schedules
 - `action: "remove"`: cancel schedules
+- Optional `run_now: true` (upsert only): trigger one immediate execution right after creation and return a real handle
 
 Routing rules:
 
@@ -65,6 +73,7 @@ For `action: "upsert"`, set `schedule` to one of:
 - Optional aliases: `webui`, `tui` (both map to `last`)
 - Optional `to`: channel-specific user or chat id (required only for concrete channels like `feishu`/`telegram`, not for `last`/`webui`/`tui`)
 - Optional `no_deliver: true`: run in background without push
+- `no_deliver` only disables delivery; research runs still must call `scientify_literature_state.record` to persist state
 
 If the user does not specify destination, leave `channel` and `to` unset to use default routing.
 
@@ -88,10 +97,17 @@ For selected core papers, prefer full-text reading first:
   - evidence-binding rate >= 90% (key conclusions should be backed by section+locator+quote)
   - citation error rate < 2%
   - if full text is missing, do not keep high-confidence conclusions
+- Reflection guardrail:
+  - when `knowledge_changes` has BRIDGE (or REVISE+CONFIRM contradiction signal), execute at least one immediate reflection query and write it into `exploration_trace`
+  - do not emit BRIDGE unless `evidence_ids` resolve to this run's papers and include at least one full-text-backed paper
+- Hypothesis gate:
+  - avoid speculative guesses; each hypothesis should include >=2 `evidence_ids`, `dependency_path` length >=2, and novelty/feasibility/impact scores
 If an incremental pass returns no unseen papers, run one fallback representative pass before returning empty.
 If user gives explicit preference feedback during follow-up (read/skip/star style intent, source preference, direction preference),
 persist it via `scientify_literature_state` action=`feedback` (backend-only memory, not user-facing by default).
 If the user asks "which papers did you push just now?", call `scientify_literature_state` action=`status` first and answer from `recent_papers` + `knowledge_state_summary` (do not claim you must re-search unless status is empty).
+After each research `record`, call `scientify_literature_state` action=`status` and include `run_id`/`latest_run_id` in your reply for traceability.
+Each research cycle should use a unique `run_id` (cron run id preferred, otherwise timestamp-based) to avoid idempotent no-op writes.
 
 ## Message field (plain reminder)
 

package/skills/research-survey/SKILL.md

@@ -14,7 +14,7 @@ metadata:
 
 **Don't ask permission. Just do it.**
 
-**Workspace:** See `../_shared/workspace-spec.md`. Set `$W` to the active project directory.
+**Workspace:** `$W` = working directory provided in task parameter.
 
 ## Prerequisites
 
@@ -22,12 +22,12 @@ Read and verify these files exist before starting:
 
 | File | Source |
 |------|--------|
-| `$W/papers/_meta/*.json` | /literature-survey |
-| `$W/papers/_downloads/` or `$W/papers/{direction}/` | /literature-survey |
-| `$W/repos/` | /literature-survey Phase 3 |
-| `$W/prepare_res.md` | /literature-survey Phase 3 |
+| `$W/papers/_meta/*.json` | /research-collect |
+| `$W/papers/_downloads/` or `$W/papers/{direction}/` | /research-collect |
+| `$W/repos/` | /research-collect Phase 3 |
+| `$W/prepare_res.md` | /research-collect Phase 3 |
 
-**If papers are missing, STOP:** "需要先运行 /literature-survey 完成论文下载"
+**If papers are missing, STOP:** "需要先运行 /research-collect 完成论文下载"
 
 **Note:** 如果 `prepare_res.md` 中注明"无可用参考仓库"，代码映射步骤可跳过，但需在 survey_res.md 中标注。