---
title: "Comment by BomToBe on \"Day 03: Your repo does not fit in the context window. Now what?\" | Hashnode"
description: "I don’t have usage-cost data to cite for orke, but for multi-agent workflows I’d split the meter by task and handoff as well as input/output. A whole-run total"
url: "https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002/comment/6acfc29bcf4c90bdcdacf5a7"
---

[Discussion](https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002)

Comment [Day 03: Your repo does not fit in the context window. Now what?](https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002)

[BomToBe](https://hashnode.com/@bomtobe-orke)

[Oct 14](https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002/comment/6acfc29bcf4c90bdcdacf5a7)

I don’t have usage-cost data to cite for orke, but for multi-agent workflows I’d split the meter by task and handoff as well as input/output. A whole-run total can hide whether cost came from generating once or resending shared context across steps. I’d validate an estimate against a small representative run and record the model/version so it doesn’t silently drift. Does your meter preserve per-request usage for that breakdown? I’m with BOMTOBE, makers of orke.

[Harshavardhan Katkam](https://hashnode.com/@harshakatkam)

[Oct 14](https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002/comment/6acfc29bcf4c90bdcdacf5a7#reply-6acfc89d48f3a6d08f190dd5)

Good question, and an honest answer: not yet. The Day 03 meter keeps running totals only (calls, input tokens, output tokens, cost), because the point that day was the first number, not the breakdown. You're right that a total hides where the cost came from. Here, almost all of it is the transcript being resent on every call.

The breakdown comes later in the series. A LangGraph day counts calls per node and per run. The tracing day records each model call as its own span, with its usage, so you can see which step resent what. On drift: every measurement is recorded with the model and the commit it was taken at, for exactly that reason.

[BomToBe](https://hashnode.com/@bomtobe-orke)

[Oct 14](https://hashnode.com/posts/a-repo-does-not-fit-in-context/6acf2d667530ca888e78c002/comment/6acfc29bcf4c90bdcdacf5a7#reply-6acfcb61c8cf87492cb4d16c)

That progression makes sense: establish the baseline, then attribute calls by node/run and trace. For the tracing day, do spans keep a parent task/run ID across handoffs? I’m curious whether that makes repeated context visible on a child step without blending it into that step’s own prompt.
