How to Check Codebuff Usage and Cost

Codebuff is a multi-agent terminal coding tool: you give it one task, and instead of answering itself, it dispatches a set of specialized sub-agents — a planner, a file editor, a reviewer, and more — that each talk to the model on your behalf. That design is great for getting work done hands-off. It is terrible for keeping a mental tally of spend, because the token cost of a single prompt is scattered across every sub-agent it triggered.

How do I see my Codebuff token usage?

One command reads the logs Codebuff already writes locally and collapses the whole agent swarm into a single, readable total:

$ npx whoburnedmore↳ scanning local Codebuff logs…↳ stitched 312 sub-agent turns into 18 tasks   CODEBUFF USAGE  ────────────────────────────────────────  2026-06-16  input  4,180,300  output  402,700  2026-06-15  input  2,905,100  output  271,400  2026-06-14  input  3,640,900  output  338,200   BY SUB-AGENT ROLE  planner      5.9M tokens  file-editor  9.4M tokens  reviewer     3.1M tokens   7-day total:  18.4M tokens   est. cost shown per model

Each line you see in your editor as “one Codebuff task” really expanded into a handful of agent conversations. whoburnedmore walks the local log entries, groups them back under the task that kicked them off, and adds up the input and output tokens — so the number you read is the truecost of the request, not just the orchestrator's slice.

What the local logs actually contain

The records carry the fields that matter for accounting: which model each sub-agent called, how many input and output tokens that call consumed, and a timestamp. whoburnedmore reads only those numeric fields — never your prompts, your code, or your file names.

1. 1

   Group sub-agent turns by parent task

   Every turn logged by a planner, editor, or reviewer is linked back to the user request that spawned it, so one task shows one honest total instead of a dozen fragments.
2. 2

   Split by model

   A fast model handling routine edits and a frontier model handling the hard reasoning are priced very differently. The breakdown keeps them separate so the estimate is real.
3. 3

   Roll up by day

   Daily totals surface the heavy days — the ones where a single ambitious task fanned out into a long chain of agent calls — instead of burying them in a monthly average.

Why is Codebuff's multi-agent cost hard to track?

The fan-out is the whole story. One thing you type becomes many model calls, and each sub-agent often re-sends shared context so it can do its job — which multiplies input tokens in a way that is invisible from the outside 🧮:

Because the orchestrator and its sub-agents each bill independently, the cost of a task is a sum you never see assembled in one place. The role breakdown makes the shape of that spend concrete — usually the editor and reviewer agents, which reread large chunks of your files, dominate:

How much is Codebuff costing me?

Cost for a multi-agent run is just the sum of every sub-agent's call, with input and output tokens priced separately per model. whoburnedmore applies the published per-token rate for each model a sub-agent used and adds them all up:

The reason this matters: a task that felt cheap because you only typed one sentence can be expensive once you count the five sub-agents it woke up. Pricing varies by tool and plan, so whoburnedmore does not guess your plan — it estimates the underlying API cost of the tokens you actually spent, which is the honest yardstick whether you are on credits or a flat fee.

One tool or all of them

If Codebuff is the only agentic tool you run, the totals are just Codebuff. If you also use Claude Code, Codex, or Gemini CLI, the same command shows them side by side — see the cross-tool usage guide for that combined view.

Is reading my Codebuff logs private?

Yes. whoburnedmore parses only token counts, model names, and timestamps from the local logs — your prompts and your source code never leave the machine. Run npx whoburnedmore --dry-run to print exactly what a submission would contain, or npx whoburnedmore --local to open a dashboard that stays entirely on your computer. 🛡️