← 返回 Skills 市场
edwardwason

Session Branch

作者 AI花生 · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ⚠ suspicious
40
总下载
1
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install session-branch
功能描述
Branch a coding session into a new conversation with full context handoff — generate structured handoff doc, startup prompts, and guide the new session to pi...
使用说明 (SKILL.md)

Session Branch

Branch your current coding session into a new conversation without losing context.

When to Use

  • Current conversation is getting long and context compression is degrading quality
  • You want to start a new task on the same project but keep full context
  • You need to fork your work into a parallel direction
  • User says: "开个支线" / "分叉" / "branch" / "新任务但保留上下文"

Execution Flow

Step 1: Analyze Current Session

Scan the current conversation and project to extract:

  1. Project identity — name, description, platforms, version, tech stack
  2. Decision chain — what was chosen, what was rejected, and why
  3. Data flow — how data moves through the system (input → process → output)
  4. Capability boundary — what the project CAN do and CANNOT do
  5. Code changes — what files were modified in this session and why
  6. Platform status — GitHub, ClawHub, npm, PyPI current state
  7. Environment — which env vars are set, which need user config
  8. Knowledge files — index of docs/knowledge/ and docs/rules/ by scenario
  9. User preferences — language, work style, code style, security sensitivity
  10. Branchable directions — what can be built next, with files involved and prerequisites

IDE-Specific Additional Scanning

For WorkBuddy, also scan:

  • Identity files: ~/.workbuddy/SOUL.md, IDENTITY.md, USER.md — persona and preferences
  • Memory files: .workbuddy/memory/MEMORY.md + daily logs — project memory
  • Installed skills: ~/.workbuddy/skills/ — list of active skills
  • Scheduled tasks: automation/cron task list and status
  • Channel config: IMA knowledge base IDs, Feishu channel configuration
  • MCP connectors: active MCP connector status

For TRAE SOLO, also scan:

  • Rules: .trae/rules/ — project-level rules
  • Schedule: TRAE SOLO Schedule task list

For Cursor, also scan:

  • Rules: .cursor/rules/ or .cursorrules

For Claude Code, also scan:

  • Rules: CLAUDE.md in project root

Step 2: Generate Handoff Document

Use the template at references/handoff-template.md to create a structured handoff document.

Save location by IDE:

IDE Default path Rationale
WorkBuddy .workbuddy/session-handoff.md Align with memory system, keep project root clean
TRAE SOLO docs/session-handoff.md Standard docs location
Cursor docs/session-handoff.md Standard docs location
Claude Code docs/session-handoff.md Standard docs location

Critical rules:

  • NO personal information (real names, emails, specific paths, token values)
  • NO project-specific secrets or credentials
  • Use generic placeholders: \x3Cproject-name>, \x3Cowner>, \x3Cyour-path>
  • The handoff doc must be reusable as a TEMPLATE, not a one-time snapshot

Step 3: Validate with Checklist

Cross-check the generated handoff against references/checklist.md.

Every item must be covered. If something is not applicable, write "N/A" with a reason.

Step 4: Generate Startup Prompt

Use references/startup-prompts.md to generate the startup prompt for the new conversation.

The prompt must include:

  1. Exact file paths for the new AI to read (absolute or relative per IDE)
  2. A three-step flow: Load → Report → Ask
  3. Clear role: "This is a continuation of an existing project, not starting from scratch"

Step 5: Present to User

Show the user:

  1. The generated handoff document (key sections summary)
  2. The startup prompt (ready to copy-paste)
  3. List of files created and their locations
  4. Ask: "Ready to open a new conversation?"

Configuration

Setting Default Description
handoff_path Auto (by IDE) Where to save the handoff document
include_checklist true Whether to validate against checklist
target_ide auto Target IDE for startup prompt (auto/trae/workbuddy/cursor/claude-code)

References

  • references/handoff-template.md — Full template for the handoff document
  • references/checklist.md — Validation checklist (12 categories + IDE-specific)
  • references/startup-prompts.md — IDE-specific startup prompt templates
安全使用建议
Install only if you are comfortable with the agent reading broad project and IDE context. Before using it, ask the agent to limit scanning to project-local files unless you explicitly approve WorkBuddy identity, memory, scheduled-task, channel, connector, and environment-variable review. Review the generated handoff file before committing or sharing it, and prefer repo-relative paths and placeholders for personal or secret values.
能力评估
Purpose & Capability
The core session-handoff purpose is coherent, and generating a handoff document plus startup prompt fits that purpose. However, the WorkBuddy-specific instructions expand into identity files, user profile, memory, installed skills, scheduled tasks, channel configuration, and MCP connector status, which is broader and more sensitive than a typical project handoff.
Instruction Scope
The skill tells the agent to scan broad project, environment, platform, preference, and IDE state, including global/home-directory WorkBuddy files. It includes privacy-filter rules, but it does not require a clear consent checkpoint before reading non-project-local personal or integration metadata.
Install Mechanism
The package consists of markdown skill and reference files with no executable scripts or dependency install behavior; static scan and VirusTotal telemetry are clean.
Credentials
Reading project files is proportionate for a handoff skill, but enumerating environment variables, platform status, global identity/memory files, scheduled tasks, channel configuration, and connector status is high-impact context collection that is only loosely bounded.
Persistence & Privilege
The skill writes a handoff document such as docs/session-handoff.md or .workbuddy/session-handoff.md and generates prompts for another AI session. This persistence is disclosed and purpose-aligned, but overwrite behavior and sensitive-content handling are not clearly controlled beyond general filtering instructions.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install session-branch
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /session-branch 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.1.0
WorkBuddy deep adaptation: identity files, memory system, skills list, scheduled tasks, IMA/Feishu channels, MCP connectors
v1.0.1
docs: update README with bilingual structure, add CONTRIBUTING.md and PUBLISHING_GUIDE.md
v1.0.0
Initial release of session-branch: branch your ongoing coding session into a new conversation with full context transfer. - Analyzes the current session to extract project identity, decision history, and context details. - Generates a reusable handoff document (no personal data or secrets) using a structured template. - Cross-checks the handoff against a validation checklist for completeness. - Creates a startup prompt for the new session with file paths and clear step-by-step instructions. - Presents the user with a summary, the handoff doc, startup prompt, file list, and confirmation to branch.
元数据
Slug session-branch
版本 1.1.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 3
常见问题

Session Branch 是什么?

Branch a coding session into a new conversation with full context handoff — generate structured handoff doc, startup prompts, and guide the new session to pi... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 40 次。

如何安装 Session Branch?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install session-branch」即可一键安装,无需额外配置。

Session Branch 是免费的吗?

是的,Session Branch 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Session Branch 支持哪些平台?

Session Branch 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Session Branch?

由 AI花生(@edwardwason)开发并维护,当前版本 v1.1.0。

💬 留言讨论