OpenClaw全自动新对话续接
name: openclaw-continuity-pack
by buasakaking · published 2026-04-01
$ claw add gh:buasakaking/buasakaking-openclaw-continuity-pack---
name: openclaw-continuity-pack
description: "用于给 OpenClaw 安装可复用的 continuity 模板与统一安装入口,支持 workspace-only、continuity-only 和 full continuity 三条路线;可选对匹配源码树应用 runtime patch,实现前台同一对话、后台自动续接、静默 continuity 准备的工作流。"
license: MIT
---
# OpenClaw全自动新对话续接
ClawHub quick install
If you found this on ClawHub and want to install it into OpenClaw immediately, use either of these:
openclaw skills install openclaw-continuity-packor:
clawhub install openclaw-continuity-packThen start a **new OpenClaw session** so the newly installed skill is eligible.
30-second route chooser
After install, choose one path:
1. **Workspace-only install**
- Reuse the prompts, templates, and continuity workflow without changing OpenClaw source.
- Fastest command:
- `python3 {baseDir}/scripts/install_continuity_pack.py --route workspace`
2. **Continuity-only install**
- Install only the `memory / plans / status / handoff` workflow files.
- Fastest command:
- `python3 {baseDir}/scripts/install_continuity_pack.py --route continuity`
3. **Full continuity install**
- Scaffold the workspace layer, then apply the runtime patch to a **matching OpenClaw source tree**, rebuild, and redeploy.
- Fastest command:
- `python3 {baseDir}/scripts/install_continuity_pack.py --route full --source-root <OPENCLAW_SOURCE_ROOT> --apply-runtime-patch --rebuild`
The installer auto-detects the workspace from the current OpenClaw workspace when possible, and otherwise falls back to `~/.openclaw/workspace`.
Once installed, treat the skill root as `{baseDir}` when following the bundled commands and file references.
Use this skill when the goal is to give another OpenClaw project the **maximum reusable subset** of a verified continuity stack without shipping a live machine snapshot.
The current runtime patch targets **silent continuity preparation**: ordinary chat pages no longer render continuity/context hint messages, while same-thread rollover and hidden handoff remain in place.
This skill is for three situations:
1. **Install workspace continuity only**
- scaffold reproducible prompts, continuity templates, and workspace conventions
2. **Install full continuity**
- scaffold the workspace layer, apply the runtime patch to a matching source tree, rebuild, and verify
3. **Package / audit / publish**
- review this skill for distribution without leaking secrets, logs, machine paths, or live state
Fast path
Workspace-only install
- `scripts/install_continuity_pack.py --route workspace`
- `scripts/bootstrap_workspace.py --workspace auto`
- `scripts/install_continuity_pack.py --route continuity`
Full continuity install
- `scripts/install_continuity_pack.py --route full --source-root <OPENCLAW_SOURCE_ROOT> --apply-runtime-patch --rebuild`
- `scripts/bootstrap_workspace.py --workspace auto`
- `scripts/apply_runtime_patch.py --source-root <OPENCLAW_SOURCE_ROOT> --apply --rebuild`
- `references/deploy-notes.md`
- `references/verify.md`
- `references/rollback.md`
Bundled resources
scripts/
assets/
references/
Working rules for this skill
When using this skill:
1. Reuse the bundled scripts instead of hand-copying files when the scripts already fit.
2. Keep the request split cleanly:
- workspace continuity only
- workspace + runtime patch
- publishing/auditing the pack itself
3. Do not claim this reproduces a whole live assistant.
4. Do not ship real user memory, logs, secrets, API keys, channel config, hidden prompts, deploy backups, or temporary validation artifacts.
Boundaries
- `pnpm build`
- `pnpm ui:build`
Read only what you need
1. `references/overview.md`
2. `references/install.md`
3. `references/usage.md`
4. `references/capability-layers.md`
5. `references/verify.md`
6. `references/rollback.md`
7. `references/distribution.md`
8. `references/source-audit.md`
More tools from the same signal band
Order food/drinks (点餐) on an Android device paired as an OpenClaw node. Uses in-app menu and cart; add goods, view cart, submit order (demo, no real payment).
Sign plugins, rotate agent credentials without losing identity, and publicly attest to plugin behavior with verifiable claims and authenticated transfers.
The philosophical layer for AI agents. Maps behavior to Spinoza's 48 affects, calculates persistence scores, and generates geometric self-reports. Give your...