kForge fits if…
- You want to use AI on your code without opening a shell to it.
- Your security team requires a single, identified access surface.
- Your repositories must stay on your servers.
kForge use case
An AI assistant on your code, without machine access
Most development assistants assume command-line access and a machine to work on. For many organisations, that is what makes AI unacceptable.
The short answer
kForge exposes 63 tools to an agent, grouped into five profiles and served by its own API. The agent browses a repository, reads the logs of a failed run or handles a code analysis finding, with no shell and no machine access.
Five profiles, limited permissions
Each profile gives access to a specific area; the overview is read-only.
- Overview: projects, repositories, pipelines, tickets, findings
- Repository: history, branches, files, diffs
- Pipelines: re-run, approve or cancel a run
- Code analysis: check findings against the code
- Tickets: create, progress, link to commits
The code stays within your perimeter
The kForge server never runs git itself: history, diffs and cloning are delegated to an agent installed on your machines.
Is it the right tool for you?
Limits to know
- The agent acts within the five profiles and the user’s permissions.
- kForge does not provide the AI model itself: it exposes its tools to the assistant of your choice.
- Pricing is still being defined.