chanakya-usage

name	chanakya-usage
description	Use when checking token burn, usage cost, or run history for chanakya agent runs

Chanakya Token Usage

When this skill is invoked, immediately run these commands and display the real output. Do not show documentation — show live data.

Step 1: List all runs

npx chanakya status

Display the output. If "No runs found" → stop, tell user no runs yet.

Step 2: Show token breakdown for each coordinator run

For each run ID from status output that has agent_started events (coordinator runs, not orchestrator):

npx chanakya tokens <run-id>

Display each output inline.

Step 3: Summarize

After showing all runs, output a summary:

--- Summary ---
Runs shown: N
Total in:   X tokens
Total out:  Y tokens
Total cached: Z tokens

Estimated cost (USD):
  gemini:      $X.XX
  claude-code: $X.XX

Cost rates:

• subagent:gemini — in: 1.25/1M, out: 10.00/1M
• subagent:claude-code — in: 3.00/1M, out: 15.00/1M
• harness — in: 3.00/1M, out: 15.00/1M

Notes

• Two runs per chanakya run: orchestrator (no LLM tokens) + coordinator (real work). Show both but flag which is which.
• chanakya usage aggregate command = Phase 19, not yet impl'd. This skill is the interim.