| name | gauge |
| description | Instant token snapshot for the current Claude Code session — context %, cost, cache efficiency, degradation risk |
Run a single token snapshot for the current Claude Code session and report back.
Steps
- Run the CLI:
python "C:/Users/albat/Desktop/Universal Brain/tools/ai-token-dashboard/delivery/cli.py" --agent cc
-
Read the output and report:
- CONTEXT % and whether it's healthy / needs compact
- Current session cost estimate
- Degradation risk level
- Advice line from the dashboard
-
If context is at 80%+ or messages are 30+, recommend running /compact immediately.
Quick Reference
- GREEN = healthy, keep going
- YELLOW = plan to compact soon
- RED = compact now
- Degradation starts at message 30+ or 80% context (Anthropic research)