Compaction Survivor Kit
SkillSkill
Your agent never loses context again — even after compaction resets everything
About
The Problem
You're deep in a conversation. Your agent has full context. Then — compaction hits. Suddenly your agent has a summary and nothing else. Key decisions, ongoing tasks, conversation tone — all gone.
The Solution: 4 Components
1. Pre-Compaction Memory Flush
OpenClaw's compaction.mode: safeguard triggers a memory flush BEFORE compacting. Your agent writes everything important to disk:
- Active task context
- Recent decisions
- Ongoing conversation state
- Pending actions
2. Post-Compaction Context Injector
After compaction, the agent automatically reads:
- Last 24h daily note summaries
- Last 15 messages from
recent-messages.md - Active task context from memory files
- MEMORY.md for long-term preferences
3. Recent Messages Tracker
A cron job (every 30 min) that saves the last 10-15 user messages to memory/recent-messages.md. After compaction, the agent reads this first to understand what was being discussed.
4. Daily Note Protocol
At end of each active session, the agent writes a daily note with:
- Key decisions made
- Tasks completed/started
- Important context for tomorrow
Config
{
"compaction": {
"mode": "safeguard",
"memoryFlush": true
}
}
Before vs. After
Before: "I don't have context about what we were discussing." After: Agent picks up exactly where you left off, references recent decisions, continues ongoing tasks.
Includes
- SKILL.md with complete protocol
- OpenClaw config patch for safeguard mode
- Cron job config for recent-messages extraction
- Daily note template
- AGENTS.md additions for recovery protocol
Core Capabilities
- Pre-compaction memory flush (automatic)
- Post-compaction context recovery
- Recent messages tracking (cron)
- Daily note protocol
- OpenClaw safeguard mode config
- Zero manual intervention required
Customer ratings
0 reviews
No ratings yet
- 5 star0
- 4 star0
- 3 star0
- 2 star0
- 1 star0
No reviews yet. Be the first buyer to share feedback.
Version History
This skill is actively maintained.
February 15, 2026
Initial release
One-time purchase
$9
By continuing, you agree to the Buyer Terms of Service.
Creator
Dima Vogel
Building AI agent systems that actually run 24/7
5 autonomous agents, 40+ cron jobs, multi-agent intelligence. Marketer turned agent architect. These skills are battle-tested in production — not weekend projects.
View creator profile →Details
- Type
- Skill
- Category
- Productivity
- Price
- $9
- Version
- 1
- License
- One-time purchase
Works With
Works with OpenClaw, Claude Projects, Custom GPTs and other instruction-friendly AI tools.
Compatible With
OpenClaw 2026.2+
Required Tools
cron
Works great with
Personas that pair well with this skill.
Agent CEO Framework
Persona
Most AI agents do what you tell them. This one decides what actually needs doing.
$49
Project Manager Agent
Persona
Goals into plans. Plans into shipped work.
$34
ClawMart QA Review Shield
Persona
Preflight ClawMart packages for originality, safety, installability, and buyer usefulness before anyone hits publish.
$24