docs: v1.2 guide on reducing token usage with AI coding agents#269
docs: v1.2 guide on reducing token usage with AI coding agents#269rachaelrenk wants to merge 11 commits into
Conversation
Co-Authored-By: Oz <oz-agent@warp.dev>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
I'm starting a first review of this pull request. You can view the conversation on Warp. I completed the review and no human review was requested for this pull request. Comment Powered by Oz |
![oz-for-oss[bot]](https://avatars.githubusercontent.com/in/3450139?s=60&v=4){kind=small}
There was a problem hiding this comment.
Overview
This PR adds a new Guides > Configuration page about reducing token and credit usage with agents, plus the corresponding sidebar entry. I checked the guide against the provided diff, existing docs references, and the security checklist; the links and referenced slash commands map to existing docs, and there are no security findings.
Concerns
- No blocking concerns found.
Verdict
Found: 0 critical, 0 important, 0 suggestions
Approve
Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).
Powered by Oz
rachaelrenk
changed the title
Incorporates the new custom model routers feature: route by complexity or rules to keep routine work off the most expensive models. Co-Authored-By: Oz <oz-agent@warp.dev>
rachaelrenk
changed the title
Credits are being phased out in favor of tokens. Removes credit framing and genericizes credit-named surfaces (usage chip, usage details, usage resets) while keeping links accurate to today's docs. Co-Authored-By: Oz <oz-agent@warp.dev>
Editorial pass from PR review: clearer intro, accurate Usage Summary description (with TODO for a token-labeled screenshot), tightened custom router and conversation sections, a Next steps lead-in, and a chose->choose typo fix. Co-Authored-By: Oz <oz-agent@warp.dev>
| @@ -0,0 +1,103 @@ | |||
| --- | |||
| title: "Reduce token usage with AI coding agents" | |||
There was a problem hiding this comment.
instead of framing it as "reduce" i would try to frame this more around how to be more efficient with tokens / make them go further / accomplish more with less for a cost-conscious audience
There was a problem hiding this comment.
@rachaelrenk I still think we could frame this more positively. Instead of “Reduce token usage,” maybe something like “Use tokens more efficiently with AI coding agents” or “Get more out of your tokens" (i defer to you on the exact messaging, but i think we should stray away from "reduce" -- we'd love to encourage folks to use our agent more!)
The guide is less about cutting usage for its own sake and more about helping cost-conscious users make their usage go further: choosing the right model, keeping context focused, using routers/rules, etc.
| --- | ||
| Every agent task consumes tokens. Tokens are the unit of text a model reads and generates. The more tokens a task uses, the more it costs and the longer it takes, so trimming token usage keeps your agent workflows lean and fast. | ||
|
|
||
| This guide covers practical ways to lower token usage in Warp. You'll learn how to choose the right model, route tasks to cost-appropriate models, keep context tight, manage conversations, and configure your agents to work efficiently. |
There was a problem hiding this comment.
it's worth mentioning somewhere that in Warp, the unit of measurement is actually credits, but it's essentially packages up tokens into an easier to grok measurement (e.g. different tokens rates across different providers, price of per token, etc)
|
|
||
| Larger reasoning models process more tokens per turn than lighter ones, so the model you choose has one of the biggest effects on usage. | ||
|
|
||
| * **Use a cost-efficient model for routine work** - Switch to **Auto (Cost-efficient)** (`auto-efficient`), which optimizes for lower token consumption while keeping output quality high. Lightweight models like Claude Haiku also use fewer tokens for simple edits, lookups, and quick questions. |
There was a problem hiding this comment.
maybe worth mentioning that if folks prefer open source models, we also have an auto-open (need to double check the exact name) router, or they can create their own custom router
There was a problem hiding this comment.
ah i see you mention the custom routers below, nvm
Per @hongyi-chen review: shift framing toward token efficiency (keeping the searchable title/slug), add a note that Warp meters usage in credits which package tokens across providers, and mention Auto (Open-weights) (auto-open) for open-source models. Co-Authored-By: Oz <oz-agent@warp.dev>
2 participants
Summary
Adds a new Guides-section guide on getting more out of your tokens with Warp's coding agents. It lives in Guides → Configuration, alongside "Use Agent Profiles efficiently."
The guide keeps the searchable title "Reduce token usage with AI coding agents" (for the non-branded search query) but frames the body around using tokens efficiently and getting more out of each token. It also notes that Warp meters this usage in credits, which package tokens into a single unit across differing provider rates.
Changes
src/content/docs/guides/configuration/how-to-reduce-token-usage-with-ai-coding-agents.mdx(new)/cost, Settings > Billing and usage)auto-efficient,auto-openfor open-weight models, lightweight models, avoiding mid-conversation model switching for prompt caching)/new,/compact,/fork-and-compact)/index)AGENTS.md(/add-rule,/init)/plan)src/sidebar.tsReview updates (@hongyi-chen)
auto-open).Notes
style_lint.pypasses with 0 issues, and a localnpm run buildsucceeds (347 pages). The failing Vercel preview is from a pre-existing/404route collision (src/pages/404.astrovs Starlight's built-in) coming frommain, not from this guide.Conversation: https://staging.warp.dev/conversation/1c22e517-4000-4f62-be81-f63070554481
Co-Authored-By: Oz oz-agent@warp.dev