Demerzels-lab Token Saver
💡 Did you know? Every time you send a prompt, your workspace files (SOUL.md, USER.md, MEMORY.md, AGENTS.md, and more) are sent along with it — every single time. These files count toward your context window, slowing down responses and costing you real money on every message.
Token Saver compresses these files using AI-efficient notation that preserves all your data while making everything lighter, faster, and cheaper.
Commands
| Command | What it does |
|---|---|
/optimize |
Full audit dashboard — files, models, compaction |
/optimize tokens |
Compress workspace files (auto-backup) |
/optimize compaction |
Chat compaction control |
/optimize compaction 120 |
Set custom compaction threshold |
/optimize revert |
Restore files from backups, disable persistent mode |
Features
📁 Workspace File Compression
Scans all .md files sent with each API call. Shows token count and potential savings per file.
File-aware compression:
- SOUL.md — Light compression, keeps evocative/personality language
- AGENTS.md — Medium compression, dense instructions
- USER.md / MEMORY.md — Heavy compression, key:value data
- PROJECTS.md — No compression (user-specific structure)
🤖 Model Audit
Detects current AI models for main chat, cron jobs, and subagents. Suggests cheaper alternatives where appropriate (e.g., Gemini free tier for background tasks).
💬 Chat Compaction
Analyzes your conversation history (default: last 7 days) to recommend optimal compaction threshold.
How it works:
- Scans your sessions to detect topic change patterns
- Recommends threshold based on YOUR usage
- Shows savings estimate per preset
Presets:
- 🟢 Safe (160K) — ~$30/mo savings
- 🟡 Balanced (120K) — ~$100/mo savings
- 🔴 Aggressive (80K) — ~$200/mo savings
Options:
--month— Scan last 30 days (more data)--all— Scan all history (takes longer)- Custom number:
/optimize compaction 100= compact at 100K tokens
📝 Persistent Mode
When enabled, adds file-type-aware writing guidance to AGENTS.md:
| File | Writing Style |
|---|---|
| SOUL.md | Evocative, personality-shaping |
| AGENTS.md | Dense instructions, symbols OK |
| USER.md | Key:value facts |
| MEMORY.md | Ultra-dense data |
| memory/*.md | Log format, dated |
| PROJECTS.md | Keep user's structure |
Safety
- Auto-backup — All modified files get
.backupextension - Integrity > Size — Never sacrifices meaning for smaller tokens
- Revert anytime —
/optimize revertrestores all backups - No external calls — All analysis runs locally
Installation
clawhub install token-saver
Version
2.0.0 — Chat compaction, file-aware compression, persistent mode