Lineman works by intercepting the data-heavy output your coding agent would otherwise pull into its context window (large files, build logs, command output, web pages) and handing it to a fast secondary model that compresses it down to what matters before it reaches your primary model. That means your code passes through our service. So it's fair to ask a plain question: what do we actually do with it?
The short answer is that we treat your code as yours. We process it to do the one job you asked us to do, and then it's gone. This post walks through how that works in plain English. The full detail lives in our Privacy Policy.
Processed in the moment, not kept
Your code is processed in real time and isn't retained after the response goes back to you. We don't keep a persistent store of your files. The compression happens, the result comes back, and the original content is discarded.
There are a few narrow operational exceptions: short-term caching to finish a multi-step request, abuse detection, security incident response, debugging a specific fault, and legal compliance. Each is subject to data minimisation and defined limits. They exist to keep the service safe and working, not to build an archive of your work, and they're spelled out in full in the Privacy Policy.
Never used to train models
This one isn't ambiguous:
Lineman does not use your code to train, fine-tune, or improve AI models. This commitment is absolute and applies to every tier and every deployment.
The data-heavy content you send through Lineman exists only to produce the summary you asked for. It is never fed back into a model as training material.
What's separate from your code
To run a reliable product we do collect some operational signals: task types, token counts, latency, error rates. That's anonymised, aggregated usage data, collected and stored separately from any code or files you process. It doesn't contain your code, and you can turn it off at any time with the /lineman:telemetry off command.
You stay in control
A few things worth knowing:
- Your work is yours. We claim no ownership over the code and files you submit.
- Telemetry is optional and can be disabled with a single command.
- Repository privacy. If you'd rather we never see your repo's remote URL, setting
LINEMAN_REDACT_REPO_REMOTE=1replaces it with an irreversible hash before anything leaves your machine.
We built Lineman to save you tokens, not to collect your code. Process what you send, return the answer, and keep nothing we don't need. If you want the precise commitments, retention rules, and your rights under UK GDPR, they're all in the Privacy Policy, and you can always reach us via contact.