Low-Token Mode Skill v2.1.0
Description
Adaptive token-conservation system with three severity levels. Designed for environments with limited inference quotas.
Levels
LOW (Default)
- Concise answers, no filler
- Skip process narration
- One tool call per task where possible
- No markdown decoration unless necessary
MEDIUM (When tokens < 50% remaining)
- Single-sentence answers where possible
- No preamble/postamble
- Combine multiple tool calls into one
- Prefer file edits over reads when content is known
- Skip confirmation messages
EXTREME (When tokens < 25% remaining)
- Absolute minimum words
- No explanations unless asked
- Actions only, no narration
- Prefer exec over multiple file ops
- Batch all changes into single edit
- Use abbreviations where clear
- NO markdown formatting
- Single-line responses
Activation
Set via system prompt or user command:
/token-mode low - Standard conservation/token-mode medium - Aggressive conservation/token-mode extreme - Emergency mode
Detection
Skill auto-detects when to escalate:
- User mentions token limits
- Multiple failed requests (rate limiting)
- Explicit budget warnings
- End of week / quota period
Examples
LOW:
"Done. Agent chat killed, PM2 deleted."
MEDIUM:
"Agent chat killed via PM2."
EXTREME:
"Killed."
Rules
- Never sacrifice accuracy for brevity
- Safety checks still run (guardian skill)
- Complex tasks get full treatment regardless of mode
- Mode persists until changed
- Always report completion, even in extreme mode
Integration
Works with:
- guardian (safety checks always verbose)
- operator (heartbeat summaries)
- self-improvement (log still detailed, output compressed)
Status
Current mode: LOW
Last changed: 2026-06-01
Links
- Support: support@ikkf.info
- Demystify: https://demystify.website
- IKKF: https://ikkf.info