You set the limits.
Decide when AI starts, where commits go, and how confident it must be.
How it works
Choose when To200 can analyze and when a human must approve.
Choose start behavior
Manual waits for your direction. Assisted can analyze and plan, then waits for approval before it starts the fix.
Control branch writes
Approve the repair before To200 writes to its isolated branch. Your default branch remains outside the launch handoff.
Choose start behavior
01Manual waits for your direction. Assisted can analyze and plan, then waits for approval before it starts the fix.
Control branch writes
02Approve the repair before To200 writes to its isolated branch. Your default branch remains outside the launch handoff.
Set confidence thresholds
03Require a minimum confidence score before pushing a branch or opening a pull request.
Escalate exceptions
Critical severity, low confidence, or missing verification can require human approval.
$ to200 policy show
launch manual + assisted
fix human-approved
merge team-controlled
✓ Check results attached to the PR
In practice
Controls teams actually need
Do not start fixes yet
Keep tasks read-only until a human moves the status forward.
Push only when confident
Require a minimum confidence score before To200 writes a branch.
Separate target branches
Route committed fixes to the branch your release process expects.
FAQ
Questions, answered.
Can we prevent To200 from starting fixes automatically?
Yes. Use manual mode so To200 waits until a human moves an issue into the fixing stage.
Can confidence be tuned?
Yes. You can set the score required before branch pushes or pull request actions are allowed.