We never see your code. By design.
A small Agent reads your code on a server you control. The answer comes back to your team — the code stays put. Built for teams whose security review is the hard part.
Your code is secure.
We never see your code. A small program — the Agent — reads it on a server you control, and the answer comes back to your team. The code itself stays put.
- ✓ The Agent runs on a server you control and reads your code right there.
- ✓ When the AI needs a few lines of code, the Agent sends them straight to the AI model — never through our servers.
- ✓ We keep only who asked, when, and how much AI it used. Never your code, never the chat itself.
- ✓ Passwords and keys found in your code are never shown in an answer.
- ✓ Need our help? Share one chat, read-only, for 1, 3 or 7 days — every view is logged.
Our cloud
Your network
Security controls that ship today.
Self-hosted Agent
A small program reads your code on a server you control. Or we run a dedicated one for you.
Direct to the model
When the AI needs a few lines, the Agent sends them straight to the model — never through our servers.
Role-aware answers
It follows the roles in your own app, so each person only hears about what they may use.
Read-only by default
Every data connection starts read-only. Changes only run inside an approved Playbook.
Audit trail
Who asked, when, and how much AI it used; every switch to Read & write is recorded.
Secrets stay secret
Passwords and keys found in your code are never shown in an answer.
Time-boxed support access
Share one chat, read-only, for 1, 3 or 7 days — every view is logged.
Customers see published content only
The Customer Assistant answers from your published Guide, never from code or data.
What our cloud sees — and what it never does.
We keep
- Who asked a question, and when
- How much AI each answer used
- Project, member and billing details
We never see
- Your source code
- The content of your chats
- Passwords and keys found in your code
Answers that know who's asking.
Your software already decides who can see what. RepoHelpdesk follows the same rules, so each person only hears about what they're allowed to use.
Yes. Open Admin → Accounts → Billing and change the renewal date. The next invoice is adjusted for the difference.
Billing dates are changed by the Billing team. Pass the request on with the customer's account number.
It reads the rules from your code
Who signs in where, the roles and permissions your code checks, and which plan unlocks what. The AI works it out — nobody writes it down.
You check it once
An admin reviews who can see what, fixes anything that is off, and tests it on a real person before switching it on.
It looks each person up
When someone signs in, RepoHelpdesk checks their role in your own system — so answers match what they can do today.
Customers too
Publish separate docs for each kind of customer. Each set only covers what those customers can use.
Security questions
Does our source code ever reach your servers?
No. The Agent reads it on your server. The few lines the AI needs go straight from the Agent to the model.
Are you SOC 2 certified?
Not yet. We'll tell you exactly where we are and walk your security team through the architecture — email security@repohelpdesk.com or book a security review.
Can we run everything inside our own network?
The Agent, the code and every database, log and API connector run inside your network. Our cloud only passes messages along and keeps usage records.
Can we fill in our own security questionnaire with you?
Yes. Send it to sales and we will complete it with you.
Bring your security team.
We'll walk them through the architecture and answer your questionnaire.