FAQ

#Getting started

#Will bmor slow down my agent?

No. bmor runs asynchronously in a detached background job, so it doesn't add any time when your agent calls its API. At session start it runs a quick login check, then does its uploading in a detached background process. If we're down or you're offline, uploads wait and go out at the next session.

#How often does bmor upload?

When a session ends, and again when the next one starts, to catch anything that didn't make it the first time. Most sessions show up in your dashboard shortly after they end. If you'd like more frequent updates, tell us on the feedback page; we're considering it.

#Should I pick Full Mode or Stats Mode?

Full Mode, unless you'd like us to collect as little as possible. It's the default, and we strongly recommend it because it powers every feature. Stats Mode works mainly as an advanced token tracker: we collect numbers and structure, none of your conversation text, and you get a subset of features. You can switch at any time. Full Mode and Stats Mode explains the difference.

#How do I uninstall?

Follow the uninstall steps on the Installation page. Uninstalling doesn't delete data you've already uploaded. Here's how to delete that.

#Security

#Can you read my code?

Not your source files. When your agent reads or edits a file, we collect the path and size, not the code, with a few exceptions. In Full Mode, though, we collect the conversation itself, so code you paste into a prompt or the agent writes in a reply is included. The full list shows exactly what we collect, and you can exclude a project to keep its code out.

#Who can see my sessions?

Only you, unless you choose to share them. Authorized staff read your data only when it's needed to fix a problem, and we review samples of sessions to improve product quality unless you opt out. Access control has the details.

#Do you train AI models on my data?

No. We don't, and neither does OpenAI, the only AI provider we send your data to. AI processing explains what we send and why.

#What happens if I paste a secret into a prompt?

bmor replaces known credential formats, like API keys and private keys, with [REDACTED] before upload. It can't recognize a password you made up, so if a real secret ended up in a session, rotate it. Secret scanning explains how it works.

#How do I report a vulnerability?

Thank you for helping keep Branchmore safe. Use the contact form to tell us what you found and how to reproduce it. Please give us a chance to fix it before you publish, and we'll credit you publicly if you'd like.

#Your data

#Can I keep some projects out of Branchmore?

Yes. List the folders to skip, or the only folders to track, in your settings file. Which projects are tracked shows how, and Exclude projects explains what we still collect.

#How long do you keep my data?

It depends on your plan. On Basics (free), we keep session logs for 3 months and metrics for 1 year. On Plus and Power user, we keep session logs for 1 year and metrics long-term. You can ask us to delete your data sooner at any time. Data retention has the full table.

#How do I delete my data?

Deleting your data also deletes your account. Send us a request, and once we've confirmed it from your account's email, we'll delete both within 30 days. Retention and deletion has the details.

#Your dashboard

#What's the difference between AI Costs and Token Costs?

AI Costs are what you actually pay for subscriptions: your Claude plan, your Cursor seat, your Copilot license. You enter these yourself.

Token Costs are estimates of what your usage would cost at published API prices. On a fixed subscription you don't pay these directly; your monthly fee covers them. If you pay API overages or use models at API prices, your real costs will track Token Costs more closely. For most people on standard plans, AI Costs is the number to budget against.

#What's the difference between rate limits and license utilization?

Rate limits are usage caps set by providers like Anthropic and OpenAI. If you've seen "rate limited, retry in X seconds," you've hit one.

License utilization is worked out from your rate-limit history: how hard you're pushing your plan compared with what it includes. High, sustained utilization suggests upgrading. Low utilization suggests you could downgrade and save money.

#Why do chat transcripts matter?

Token counts tell you what a session cost. The transcript tells you why. It's the record of what your agent actually did, which is what you need the day a file disappears, a secret gets committed, or a deploy goes wrong. It also makes the numbers useful: where the model pushed back on you, which tool calls deserve a second look, and why one session cost eight times as much as the others.

Your agent also deletes its own local logs over time; Claude Code's default is 30 days. Branchmore keeps them for your plan's retention period: 3 months on Basics, 1 year on Plus and Power user.

#What is Tool Call Safety?

Coding agents run real commands on your machine: they write files, run shell commands, and call APIs. Usually that's what you want, but some mistakes can't be undone. Tool Call Safety collects your agents' tool calls across sessions and flags the ones worth reviewing, such as destructive-looking commands, access to sensitive paths, or unusual bursts of shell activity. It needs Full Mode.

#What is Disagreement Detection?

It finds sessions where the agent pushed back on a request, where you had to override what it produced, or where you went back and forth without agreeing. A high rate usually means your context files (CLAUDE.md, AGENTS.md), your prompts, or how you scope requests could improve. Fewer disagreements save you time and tokens. It's in early access and needs Full Mode.

#Company and pricing

#Who's behind Branchmore?

Andrew Shu, the founder, has spent 16 years as an engineer and engineering leader at Permiso Security, Cisco Meraki, and Printivity. Branchmore is bootstrapped. More about us.

#What does it cost?

The Basics plan is free. Plus and Power user are paid plans, but both are free while we're in beta. See plans.

#Can my team use Branchmore?

Accounts are individual today. If you'd like to use Branchmore across a team, get in touch. We're working with early teams on what that should look like.

#How do I get help or suggest a feature?

For help, contact support. For security issues, contact security. For feature ideas, use the feedback page in the app. For quick questions, ask on Discord.