session-guardian
Never lose a conversation again. Auto-backup, smart recovery, and health monitoring for OpenClaw sessions. Protects against gateway crashes, model disconnections, and token overflow. Use this skill when: - User worries about losing conversations after gateway restart or model crash - User mentions session backup, conversation recovery, session protection, or data loss - User's agent is slow or timing out (likely token overflow from large sessions) - User runs multiple agents and needs to track collaboration across sessions - User asks about session health, backup strategy, or disaster recovery - User mentions "对话丢失", "会话备份", "上下文溢出", "token超限", "Gateway重启后记忆丢失" - Even if user just says "my agent lost everything after a restart" — this is the skill
About this skill
Session Guardian is a critical utility skill designed to ensure the persistence and integrity of conversations within the OpenClaw AI agent environment. It actively safeguards against common pitfalls such as gateway crashes, model disconnections, and token overflow, which can otherwise lead to irreversible loss of conversation history and context. The skill offers automated backup mechanisms, including regular snapshots, and provides a streamlined "one-command recovery" process to quickly restore sessions to a previous stable state. This skill is essential for users who rely on long-running or complex AI conversations, ensuring that valuable progress and context are never lost. It operates as a protective layer, continuously monitoring session health and proactively addressing potential issues like slow performance due to excessive token usage by detecting and alerting on token overflow. By integrating Session Guardian, users can mitigate the risks associated with volatile agent environments, maintaining continuity and efficiency across their AI interactions.
Best use case
The primary use case is to prevent data loss and ensure operational continuity for AI agents, especially for users engaged in extended, critical, or collaborative conversations within the OpenClaw ecosystem. It benefits anyone who needs to safeguard their AI agent's memory and context against unexpected system failures or resource exhaustion, making it indispensable for maintaining productivity and trust in their AI interactions.
Never lose a conversation again. Auto-backup, smart recovery, and health monitoring for OpenClaw sessions. Protects against gateway crashes, model disconnections, and token overflow. Use this skill when: - User worries about losing conversations after gateway restart or model crash - User mentions session backup, conversation recovery, session protection, or data loss - User's agent is slow or timing out (likely token overflow from large sessions) - User runs multiple agents and needs to track collaboration across sessions - User asks about session health, backup strategy, or disaster recovery - User mentions "对话丢失", "会话备份", "上下文溢出", "token超限", "Gateway重启后记忆丢失" - Even if user just says "my agent lost everything after a restart" — this is the skill
Users should expect their OpenClaw AI conversations to be automatically backed up, quickly recoverable after an incident, and proactively monitored for health issues like token overflow.
Practical example
Example input
My agent just crashed and I lost our entire conversation. Can you help me recover it?
Example output
Session Guardian detected a crash and successfully restored your conversation from the last hourly snapshot. All progress recovered. Your current session is now protected.
When to use this skill
- When concerned about losing conversation history due to agent crashes, restarts, or model disconnections.
- To ensure automatic backup and smart recovery for critical OpenClaw sessions.
- If an agent experiences slowdowns or timeouts, potentially due to token overflow from large conversations.
- To monitor session health, track collaboration across multiple agents, and implement a disaster recovery strategy for AI conversations.
When not to use this skill
- If you are not using an OpenClaw AI agent environment.
- For very short, non-critical, or single-turn interactions where session persistence is irrelevant.
- If your current agent platform already provides robust, built-in session backup and recovery that meets your needs.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/session-guardian/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How session-guardian Compares
| Feature / Agent | session-guardian | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | easy | N/A |
Frequently Asked Questions
What does this skill do?
Never lose a conversation again. Auto-backup, smart recovery, and health monitoring for OpenClaw sessions. Protects against gateway crashes, model disconnections, and token overflow. Use this skill when: - User worries about losing conversations after gateway restart or model crash - User mentions session backup, conversation recovery, session protection, or data loss - User's agent is slow or timing out (likely token overflow from large sessions) - User runs multiple agents and needs to track collaboration across sessions - User asks about session health, backup strategy, or disaster recovery - User mentions "对话丢失", "会话备份", "上下文溢出", "token超限", "Gateway重启后记忆丢失" - Even if user just says "my agent lost everything after a restart" — this is the skill
How difficult is it to install?
The installation complexity is rated as easy. You can find the installation instructions above.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
Related Guides
Top AI Agents for Productivity
See the top AI agent skills for productivity, workflow automation, operational systems, documentation, and everyday task execution.
Best AI Skills for ChatGPT
Find the best AI skills to adapt into ChatGPT workflows for research, writing, summarization, planning, and repeatable assistant tasks.
AI Agents for Freelancers
Browse AI agent skills for freelancers handling client research, proposals, outreach, delivery systems, documentation, and repeatable admin work.
SKILL.md Source
# Session Guardian 🛡️
Never lose a conversation again.
```
Without Guardian With Guardian
┌──────────────────┐ ┌──────────────────────────┐
│ Gateway crashes │ │ Auto-backup every 5 min │
│ → conversation │ │ Hourly snapshots │
│ gone forever │ │ Health monitoring │
│ │ │ One-command recovery │
│ Token overflow │ │ │
│ → agent frozen │ │ Gateway crash? │
│ │ │ → Restore in seconds │
│ No way back. 😱 │ │ │
└──────────────────┘ │ Token overflow? │
│ → Auto-detected + alert │
│ │
│ Always protected. 🛡️ │
└──────────────────────────┘
```
## The Problem
Your OpenClaw conversations live in session files. When things go wrong — and they will — you lose everything:
- 🔴 **Gateway restart** → entire conversation history gone
- 🔴 **Model disconnection** → mid-task context wiped out
- 🔴 **Token overflow** → session too large, agent slows down or crashes
- 🔴 **Disk issues** → session files corrupted or lost
No built-in backup. No recovery. No warning before it's too late.
## The Fix
```bash
clawhub install session-guardian
bash ~/.openclaw/workspace/skills/session-guardian/scripts/install.sh
```
That's it. Five layers of protection, running automatically.
## How It Works
```
Layer 1: Incremental Backup → Every 5 min → New conversations saved
Layer 2: Hourly Snapshot → Every hour → Full session snapshots
Layer 3: Health Check → Every 6 hours → Detect problems early
Layer 4: Smart Summary → Daily → Key info extracted
Layer 5: Knowledge Extraction → Daily → Best practices saved
```
All automatic. All in the background. Zero manual work.
## Quick Start
```bash
# Install
clawhub install session-guardian
# Deploy (sets up all cron jobs automatically)
cd ~/.openclaw/workspace/skills/session-guardian
bash scripts/install.sh
# Check status
bash scripts/status.sh
```
## What You Get
- ✅ **Auto-backup** every 5 minutes — never lose more than 5 min of conversation
- ✅ **Hourly snapshots** — full point-in-time recovery
- ✅ **Health monitoring** — warns before token overflow crashes your agent
- ✅ **One-command recovery** — restore any session from any backup
- ✅ **Multi-agent support** — protects all your agents, not just main
- ✅ **Collaboration tracking** — see task flow across agents (King → Lead → Members)
- ✅ **Knowledge extraction** — auto-saves best practices from conversations
- ✅ **Minimal overhead** — ~10-15k tokens/day, ~1-2MB/agent/month
## Usage
### Check Status
```bash
bash ~/.openclaw/workspace/skills/session-guardian/scripts/status.sh
```
Shows: backup count, last backup time, snapshot count, cron jobs, disk usage.
### Run Health Check
```bash
bash ~/.openclaw/workspace/skills/session-guardian/scripts/health-check.sh
```
Detects: oversized sessions, missing configs, disk space issues, gateway problems.
### Restore a Session
```bash
# List available backups
bash ~/.openclaw/workspace/skills/session-guardian/scripts/restore.sh list
# Restore specific agent
bash ~/.openclaw/workspace/skills/session-guardian/scripts/restore.sh restore --agent main
```
### View Collaboration Health (Multi-Agent)
```bash
bash ~/.openclaw/workspace/skills/session-guardian/scripts/collaboration-health.sh report
```
### Track Task Flow
```bash
bash ~/.openclaw/workspace/skills/session-guardian/scripts/collaboration-tracker.sh trace "task name"
```
## Cron Jobs (Auto-Configured)
After `install.sh`, these run automatically:
| Schedule | Task | What it does |
|----------|------|-------------|
| Every 5 min | Incremental backup | Save new conversations |
| Every hour | Snapshot | Full session archive |
| Every 6 hours | Health check | Detect problems |
| Daily 2am | Smart summary | Extract key info |
## File Structure
```
Assets/SessionBackups/
├── incremental/ # Every-5-min backups
├── hourly/ # Hourly snapshots
├── collaboration/ # Task flow records
└── Knowledge/ # Extracted best practices
```
All backups stay local. Nothing leaves your machine.
## Who Is This For
### Solo Agent User
✅ Protect your main conversation from crashes
✅ Get warned before token overflow kills your agent
✅ Recover in seconds, not hours of re-explaining
### Multi-Agent Team
✅ Protect all agents (King + team leads + members)
✅ Track collaboration: who assigned what, who finished what
✅ Health scoring: monitor team communication quality
✅ Knowledge extraction: auto-save best practices from every agent
## Troubleshooting
| Problem | Fix |
|---------|-----|
| Backups not running | Check `crontab -l \| grep session-guardian` |
| Agent slow/timing out | Run health-check — likely token overflow |
| Can't restore | Run `restore.sh list` to see available backups |
| Disk filling up | Check config retention settings (default: 7 days incremental, 30 days snapshots) |
## Configuration
Edit `scripts/config.sh` to customize:
```bash
BACKUP_DIR # Where backups go (default: Assets/SessionBackups)
INCREMENTAL_KEEP # Days to keep incremental backups (default: 7)
SNAPSHOT_KEEP # Days to keep snapshots (default: 30)
MAX_SESSION_SIZE # Alert threshold for session size (default: 5MB)
```
## Security
- All data stays local — no external services, no network requests
- No API keys required
- Backups are excluded from git (add to .gitignore)
- Does not modify your session files — only reads and copies
## Performance
| Operation | Tokens | Storage |
|-----------|--------|---------|
| Incremental backup | ~100/run | ~10KB/session/day |
| Hourly snapshot | ~500/run | ~100KB/session/day |
| Health check | ~200/run | ~2KB/report |
| Daily summary | ~5k/run | ~5KB/day |
| **Total** | **~10-15k/day** | **~1-2MB/agent/month** |
## Feedback
- Star: `clawhub star session-guardian`
- Update: `clawhub update session-guardian`
---
# 🇨🇳 中文说明
## Session Guardian 🛡️ — 再也不丢对话
你的 OpenClaw 对话存在 session 文件里。Gateway 一重启、模型一断连、token 一溢出——对话就没了。没有备份，没有恢复，没有预警。
## 痛点
- 🔴 **Gateway 重启** → 整个对话历史消失
- 🔴 **模型断连** → 做到一半的任务上下文全丢
- 🔴 **Token 溢出** → session 太大，Agent 变慢甚至崩溃
- 🔴 **磁盘问题** → session 文件损坏或丢失
## 解决方案
```bash
clawhub install session-guardian
bash ~/.openclaw/workspace/skills/session-guardian/scripts/install.sh
```
五层自动防护，后台运行，零手动操作。
## 五层防护
| 层级 | 频率 | 作用 |
|------|------|------|
| 增量备份 | 每 5 分钟 | 保存新对话，最多丢 5 分钟 |
| 快照备份 | 每小时 | 完整 session 存档，支持回滚 |
| 健康检查 | 每 6 小时 | 提前发现问题（token 溢出、磁盘不足） |
| 智能总结 | 每天 | 提取关键信息到 MEMORY.md |
| 知识沉淀 | 每天 | 自动保存最佳实践 |
## 你会得到
- ✅ **自动备份** — 每 5 分钟一次，最多丢 5 分钟对话
- ✅ **一键恢复** — 从任意备份点恢复 session
- ✅ **健康监控** — 在 token 溢出前预警
- ✅ **多 Agent 支持** — 保护所有 Agent，不只是 main
- ✅ **协作追踪** — 可视化任务流转（King → 团长 → 成员）
- ✅ **知识沉淀** — 自动从对话中提取最佳实践
- ✅ **极低开销** — 每天约 10-15k tokens，每月约 1-2MB/Agent
## 常用命令
```bash
# 查看状态
bash scripts/status.sh
# 健康检查
bash scripts/health-check.sh
# 恢复 session
bash scripts/restore.sh list
bash scripts/restore.sh restore --agent main
# 协作健康度（多 Agent）
bash scripts/collaboration-health.sh report
```
## 适用场景
**个人用户**：保护主对话不丢失，token 溢出前预警，秒级恢复
**多 Agent 团队**：保护所有 Agent，追踪协作链路，自动沉淀知识
## 安全性
- 所有数据本地存储，不联网，不需要 API Key
- 不修改 session 文件，只读取和复制
- 备份文件已排除 git 追踪Related Skills
find-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
filesystem
Advanced filesystem operations for listing files, searching content, batch processing, and directory analysis. Supports recursive search, file type filtering, size analysis, and batch operations like copy/move/delete. Use when you need to: list directory contents, search for files by name or content, analyze directory structures, perform batch file operations, or analyze file sizes and distribution.
Budget & Expense Tracker — AI Agent Financial Command Center
Track every dollar, enforce budgets, spot spending patterns, and build wealth — all through natural conversation with your AI agent.
yt-dlp
A robust CLI wrapper for yt-dlp to download videos, playlists, and audio from YouTube and thousands of other sites. Supports format selection, quality control, metadata embedding, and cookie authentication.
time-checker
Check accurate current time, date, and timezone information for any location worldwide using time.is. Use when the user asks "what time is it in X", "current time in Y", or needs to verify timezone offsets.
pihole-ctl
Manage and monitor local Pi-hole instance. Query FTL database for statistics (blocked ads, top clients) and control service via CLI. Use when user asks "how many ads blocked", "pihole status", or "update gravity".
mermaid-architect
Generate beautiful, hand-drawn Mermaid diagrams with robust syntax (quoted labels, ELK layout). Use this skill when the user asks for "diagram", "flowchart", "sequence diagram", or "visualize this process".
memory-cache
High-performance temporary storage system using Redis. Supports namespaced keys (mema:*), TTL management, and session context caching. Use for: (1) Saving agent state, (2) Caching API results, (3) Sharing data between sub-agents.
mema
Mema's personal brain - SQLite metadata index for documents and Redis short-term context buffer. Use for organizing workspace knowledge paths and managing ephemeral session state.
file-organizer-skill
Organize files in directories by grouping them into folders based on their extensions or date. Includes Dry-Run, Recursive, and Undo capabilities.
media-compress
Compress and convert images and videos using ffmpeg. Use when the user wants to reduce file size, change format, resize, or optimize media files. Handles common formats like JPG, PNG, WebP, MP4, MOV, WebM. Triggers on phrases like "compress image", "compress video", "reduce file size", "convert to webp/mp4", "resize image", "make image smaller", "batch compress", "optimize media".
edge-tts
Text-to-speech conversion using node-edge-tts npm package for generating audio from text. Supports multiple voices, languages, speed adjustment, pitch control, and subtitle generation. Use when: (1) User requests audio/voice output with the "tts" trigger or keyword. (2) Content needs to be spoken rather than read (multitasking, accessibility, driving, cooking). (3) User wants a specific voice, speed, pitch, or format for TTS output.