Frequently asked questions

What Lineman is, how it saves tokens, and how to get started.

What is Lineman?

Lineman is a Claude Code plugin that hands data-heavy work to a fast secondary AI: the large file reads, build and test log triage, and web-search and grep output. That keeps your primary model's context focused on reasoning, and teams typically save 40%+ tokens per session.

How does Lineman save tokens?

It intercepts the high-volume tool calls in your agent session (file reads, command output, search results) and returns a compact summary instead of the raw output. Your primary model is the expensive one, and it ends up reading far less, so the same task costs significantly fewer tokens.

Does Lineman work with Claude Code?

Yes. Lineman installs as a plugin in Claude Code and works with MCP-compatible AI coding agents. It runs alongside your normal workflow, with no change to how you prompt.

Will Lineman reduce my output quality?

No. Lineman compresses the data your model reads, not its reasoning. The model still does the thinking; it just isn't forced to read megabytes of raw logs or files first. Our published benchmarks show output quality holds while token usage drops.

How do I install Lineman?

Run `npx -y @lineman-io/cli@latest setup` in your terminal. One command installs the Claude Code plugin, installs the lineman CLI, and signs you in. Then use `claude` as you normally would, or run `lineman` instead for automatic context compaction. The Get Started guide walks through it in a couple of minutes.

Is my code safe?

Code and prompts that pass through Lineman are not used to train any model, and all traffic is encrypted in transit. See the Security and Privacy pages for the full data-handling details.

Who is Lineman for?

Individual developers and engineering teams using Claude Code who want to cut token spend without losing output quality.

Why is my Claude Code bill so high?

Most of a Claude Code bill is tool output: file reads, build and test logs, and search results loaded into context, not the model's reasoning. On Lineman's benchmarks that's over half a typical bill, and because every token in context is re-billed each turn, the same output is paid for repeatedly. Our guide on why your Claude Code bill is so high explains how to diagnose it.

How much does Claude Code cost per developer?

Anthropic's own figure is roughly $13 per developer per active day and $150–250 per developer per month across enterprise deployments, with costs under $30 a day for 90% of users (per Anthropic's Claude Code cost docs, code.claude.com/docs/en/costs). Heavy use on large repositories trends higher, largely because of how much tool output enters context.

What is context compounding?

Context compounding is the way AI coding costs grow over a session: every turn re-sends the whole conversation, including all earlier tool output, as input tokens. Anything bulky you read once is re-billed on every later turn, so cost climbs with conversation length even when your prompts stay short.

Why does tool output use so many tokens?

Tool output is large, low-signal, and sticky. A single file read or test log can be thousands of tokens the model never needed in full, and once it's in context it's re-billed every turn until you clear or compact. That's why it's over half of a typical bill, and why compressing it is where Lineman saves the most.

How do I reduce Claude Code token usage?

Route easy work to a cheaper model, run /clear at task boundaries, /compact long sessions, keep CLAUDE.md lean, and scope your file reads. The largest single lever is compressing tool output before it reaches the model, which Lineman does automatically with no workflow change. See our guide on ways to cut Claude Code token costs.

Can I reduce token costs without changing my workflow?

Yes. The manual tactics need discipline every session, but compressing tool output before it reaches the model happens automatically: you keep prompting exactly as you do now while the largest cost is cut. That's the lane Lineman occupies: 40%+ fewer tokens on our benchmarks, with output quality held.

Still have a question? Contact us.