The repository is part of the workspace
Connect one repo and the loop closes in both directions: what your team ships flows into the AI’s context as recent commits, and what the AI produces goes back out as pull requests and skills — reviewed, versioned and yours.
Your tools know about the plan, never the code
The tracker knows what you meant to do. The repository knows what you actually did. Nothing in between reads both, so the AI you ask about the project is reasoning from a description of the work rather than the work.
“Is that done?” is unanswerable
The ticket says In Progress. The branch says merged on Tuesday. Only a human closes that gap.
Integrations are one-way
A commit posts a message into a channel and that is the end of it. Nothing reads it afterwards.
AI output has nowhere to go
Suggestions in a chat window are not reviewable, not versioned, and not anyone’s to own.
How it works
Connect the repo
Settings → GitHub, through a GitHub App with the narrow permissions it needs. One workspace, one repository.
Commits become context
Recent history is part of what the AI reads on every request, so it answers about what actually shipped.
Work comes back as pull requests
Coding agents, Lab and proposed skills all end in a PR on your repo. Your review is the gate — every time.
The specifics
Scoped to the bound repo
The token an agent run receives is minted for that one repository with minimal permissions, never the whole installation.
Skills live in the repo too
What your agents learn is a file under `.decisive/skills/`, versioned and reviewable like everything else.
Free to connect
Connecting a repository and running coding agents are on every plan. Lab and Apps are the paid part.
Next to it in the product
All features
Name your team. The AI already clocked in.
Free for up to 10 people, with 1,000 AI credits included. No credit card, no sales call, no onboarding deck.