What the pilot is
A free 30-day trial of Mockingbyrd on three to ten real developer machines. Nothing to sign up for. What we collect, and what we never do, is listed below.
- Kickoff, 30 to 45 minutes. We run the coverage scan together, live, on your own machine.
- Weeks 1 to 3. You apply the highest-severity fixes, each a script you read first, and turn on the guard.
- Day 3 and day 14. Two short check-ins, so broken things get fixed fast.
- Wrap-up. We rerun the scan, compare it with your starting point, and decide together. No obligation.
What you get
- A free 30-day pilot, a direct line to the founding team, and a say in what we build next.
- A coverage scan at the start and the end, with the evidence behind every point.
- We will also ask how much time you spend managing agent permissions, before and after, and share what you tell us. It is your own estimate, compared with your own baseline.
- Afterwards, team seats are about $25 per developer per month. There is no obligation to buy.
What data we collect
Mockingbyrd reads your agent settings and session files on each machine to score, fix and guard. That reading happens on the device. During the pilot, this is everything that reaches us:
- Your survey answers, the email address you give, and any name or company you add. They are stored on our own server and read by the founding team.
- Reports you choose to send us. At the start, the end and the check-ins, you can export a scan report and a Token Ledger report. You read them first and decide what to send. A scan report includes the evidence behind each finding, which can name settings and file paths, so look before you share.
- The app's own network calls, all listed in Settings: an update check (app version, macOS version and IP address), the download, and a crash report only if you turn crash reporting on.
- Only if your team later chooses to join a workspace (not available yet): each enrolled machine would send pass or fail status per rule, scores and the pack version. Never evidence text, file contents, paths, allow-list entries, account names, or what an agent did. The app shows you the exact report before it is sent, and you can leave at any time.
We never receive your session transcripts, your files, your prompts or your tokens. The Privacy Policy has the full list and how long we keep each item.
Is it a fit?
Supported today. Apple silicon machines on macOS 14 or later. The guard enforces on Claude Code; the roster and Token Ledger also read Codex.
We ask of you. A named sponsor, three to ten developers for 30 days, the two check-ins, and a short review at the end.
Not yet. Windows or Linux, a central admin console or single sign-on, and custom builds.
Start with the fit survey
About four minutes, mostly choices. It tells us whether the pilot fits your setup today, so the kickoff is a live scan on your device and not a sales call. We will email you within a day.
The survey runs on our own server at pilots.mockingbyrd.io; your answers are stored there and go to the founding team only. Questions: hello@mockingbyrd.io · Privacy Policy