Blog
Notes on measuring what coding agents cost a team · RSS
What Claude Code, Codex and the other agents spend, per engineer, team, model and repo, and how to know it without reading anyone's prompts.
Written by the person building Kibble. The technical claims are the ones the product depends on; where a post says how a file is laid out or how a rate works, the collector's source is the thing to check it against.
-
How to track GitHub Copilot usage across a team
What GitHub's usage metrics API reports since the June 2026 move to AI credits, what a local collector adds from Copilot CLI and VS Code by model and repository, and what stays unknown.
-
How much does GitHub Copilot cost per engineer per month?
Since June 2026 a Copilot seat includes AI credits and the rest is token-priced. Three labelled scenarios at GitHub's per-model rates, and where a Business seat's $19 allowance runs out.
-
GitHub Copilot CLI logs: locations and VS Code history
Find Copilot CLI logs in ~/.copilot/session-state and ~/.copilot/logs, plus VS Code chat history paths on Windows, macOS and Linux. Count tokens safely.
ccusage for a team: what it covers and where it stops
Compare ccusage, community team tools and Kibble for Claude Code and Codex: local estimates, shared reporting, privacy, setup and billed reconciliation.
Codex usage dashboard: check limits, tokens and team usage
Check Codex usage with /status, /usage and OpenAI's dashboard. Track tokens across a team's machines, with clear limits on cost estimates and coverage.
Which Claude Code skills does your team actually use?
Measure Claude Code skill adoption with inventory and observed invocations. A worked cleanup example, context-estimate limits and why Codex zeros do not mean unused.
-
How much does Claude Code cost per engineer per month?
Illustrative Claude Code API-cost scenarios: about $5, $80 and $770 a month on Sonnet 5. Full token assumptions and dated rates, not measured team averages.
-
Why did our coding-agent bill double this week?
Six causes, in the order to check them: headcount, one engineer going heavy, a model mix shift, a falling cache hit rate, a runaway loop, or last week arriving late. The fingerprint each one leaves, and how to tell real growth from an accounting artifact.
-
Kibble vs LiteLLM, Helicone and Portkey for coding-agent cost
Gateways, ccusage and Kibble compared for coding-agent spend per engineer: what each sees, what each costs, and who shows the figure the vendor actually billed.
-
How to monitor, evaluate and improve an engineering team in the AI era
A CTO's working set of measures for a team that ships with coding agents: delivery, spend per engineer and repo, and the skills that actually fire, and what to change when each one moves.
-
How to see Claude Code cost per engineer, without reading anyone's prompts
Three ways to break one Anthropic invoice into spend per engineer, team, model and repo: a local usage tool, a proxy, or a collector that reads the logs the agent already writes and sends counts only. What each one sees, and what it costs you in privacy.
-
Where Claude Code and Codex keep their usage logs, and what a token count is worth
The JSONL transcripts under
~/.claude/projectsand~/.codex/sessions, the four token fields in them, why cache reads dominate the tokens but not the cost, and which fields a collector should never send. -
Estimated vs billed: why a coding-agent cost dashboard must never add the two
Local logs give you tokens at list price; the vendor's Admin API gives you what was charged. They disagree for six good reasons. The rule that keeps a dashboard honest: label both, let billed win the day, never sum them.