Repository
Which products and codebases are driving AI-assisted work and spend?
Claude Code analytics
Who is using AI, where,
and what does it cost?
Claude Code analytics turns raw token usage into attributed, cost-aware telemetry: usage by repository, developer, team, and organization; model-aware estimated USD cost; and trends you can explore over time.
Developer
Who is adopting the tools, and how does usage change over time?
Team
Where are workflows taking hold—and where might enablement help?
Organization
What is the total footprint across tools, accounts, and repositories?
Analytics, not a meter
A total answers “how much.” Attribution answers “why.”
A token count is a useful starting point. A team needs the surrounding facts: the model, tool, account, developer, and repository involved; the mix of input, output, and prompt-cache usage; and the price of each category.
With that context, engineering leaders can distinguish broad adoption from one unusual workload, spot changing cost patterns, and make enablement or budget decisions using the same dimensions they use for the rest of engineering.
Cost-enriched OTLP
The destination can still be yours.
Forward attributed, priced usage as OpenTelemetry to Datadog, Grafana, Honeycomb, or another OTLP-compatible destination. Build alerts, dashboards, and correlations next to the delivery and reliability data your organization already owns.
TokenMaxing provides the hosted rollups and enrichment; your observability stack remains a first-class place to work with the result.
Claude Code + Codex
Session usage
TokenMaxing
Attribution + cost
Datadog · Grafana · Honeycomb
Your OTLP destination
One view, both tools
Codex belongs in the same analysis.
Teams rarely standardize perfectly on one coding agent. TokenMaxing normalizes Codex token usage and cost alongside Claude Code, so repository and developer rollups reflect the work rather than the vendor boundary. For the details of counting each token category, read the Claude token counter guide.
Start with one collector
From install to the first team rollup.
- 1
Connect
Install the desktop app and connect the Claude Code or Codex sessions on your machine.
- 2
Sync
When enabled, normalized usage metadata is attributed and synced; raw logs remain local.
- 3
Roll up
Invite the team to see private repo, developer, team, and organization trends together.
Want a local-only check first? See the honest TokenMaxing vs ccusage comparison.
Frequently asked questions
Claude Code analytics FAQ
What is Claude Code analytics?
Claude Code analytics turns session-level token usage into cost-aware telemetry. It shows usage by model, repository, developer, team, and organization, including input, output, cache-read, and cache-write tokens and their estimated USD cost.
How do I track Claude Code cost by team?
Install TokenMaxing for each developer, enable cloud sync, and organize members into a team. TokenMaxing sends normalized usage metadata—not raw logs, prompts, transcripts, or source code—and rolls the model-aware cost up by developer, repository, team, and organization.
Can I forward Claude Code analytics to Datadog?
Yes. TokenMaxing can forward attributed, cost-enriched Claude Code and Codex usage as OpenTelemetry to Datadog or another OTLP-compatible destination, including Grafana and Honeycomb.
Get cost-attributed Claude Code analytics.
Start with your own sessions, then roll usage up across repositories, developers, teams, and the organization—or forward the enriched telemetry to your existing stack.