AgentMeter tracks token usage and cost for every AI agent run in GitHub Actions — and posts a cost summary directly on the pull request.

Works with Claude Code, Codex, Cursor, and GitHub Agentic Workflows.

No infrastructure. No config files. Three lines of YAML.

• Per-run cost broken down by input, output, and cache tokens
• Full run history and spend trends at agentmeter.app
• Free for individual repos — Pro for teams

What is AgentMeter?

AgentMeter gives developers and teams full visibility into AI agent runs — what triggered each run, what it cost, and whether it succeeded.

Every time an agent workflow completes, AgentMeter records token counts, calculates cost, and posts a summary on the triggering PR or issue. All runs are available in the dashboard at agentmeter.app.

Features

Cost tracking per run — broken down by input, output, cache read, and cache write tokens.

PR and issue comments — cost summary posted automatically on the triggering PR or issue.

Run history dashboard — filter by repo, engine, and status. Sort by cost or time.

Spend trends — track usage over time. Spot expensive weeks before they hit your API bill.

Team support — invite members to share visibility across the org.

How it works

1. Install the GitHub App
2. Add the Action — three lines of YAML at the end of your agent workflow
3. AgentMeter records every run, calculates cost, and posts a PR comment automatically

Pricing

Free — 1 repo, 30-day retention, full run history, PR comments

Pro ($19/month) — unlimited repos, 90-day retention, trends, budget alerts, team members

Full docs at github.com/agentmeter/agentmeter-action