agent-cost-monitor
by neal-collab
Coding Agents & IDEsBrief
Real-time token usage and cost tracking across all your OpenClaw agents — alerts, budgets, and optimization tips.
Description
What This Skill Does
Tracks token usage and estimated costs across multiple OpenClaw agents using the built-in session_status tool. Compares totals against configurable daily and weekly budget thresholds, then surfaces model-switching and configuration changes to reduce spend.
Because it uses OpenClaw's built-in session_status tool, the monitoring itself adds zero additional API cost.
When to Use It
- Checking which agent burned the most tokens overnight
- Forecasting whether you'll hit a weekly rate limit before Friday
- Running an automated daily cost digest via cron at a set hour
- Spotting whether heartbeat pings are using an expensive model unnecessarily
- Getting a projected weekly spend before approving a new agent workflow
Example Workflow
User asks: Give me a cost report for all agents
- 1
Calls session_status for each active agent to retrieve token counts
- 2
Calculates per-agent estimated costs using the model pricing reference table
- 3
Compares each agent's daily total against the configured per-agent max
- 4
Flags any agents or weekly projections that exceed budget thresholds
- 5
Generates a markdown report with a per-agent breakdown and ranked optimization recommendations
Markdown table showing per-agent token counts, estimated costs, budget warnings, and specific config changes to reduce spend (model downgrade suggestions, heartbeat interval, cache settings)
Installation
clawhub install neal-collab/agent-cost-monitorRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.