Parallel AI Development

Parallel AI development turns sequential feature work into a concurrent operation. Instead of one developer working on one branch, you describe the features you need and Stoneforge lets you run multiple AI agents simultaneously — each building, testing, and merging a separate feature.

Isolated worktrees, zero interference

Each agent gets its own git worktree — a fully isolated copy of your repository. Agent-1 can be building a new API endpoint while Agent-2 rewrites the dashboard component and Agent-3 adds email notifications. They never step on each other’s code.

# Create a plan with multiple features
sf plan create --title "Sprint 14 Features"

# Add feature tasks — they'll be dispatched in parallel
sf task create --title "Add user invitation flow" --plan "Sprint 14 Features"
sf task create --title "Build analytics dashboard" --plan "Sprint 14 Features"
sf task create --title "Implement webhook retry logic" --plan "Sprint 14 Features"

# Start orchestration — agents pick up tasks automatically
sf serve

Dependency-aware task dispatch

Not all features are independent. Stoneforge’s dependency system lets you express relationships between tasks. If the analytics dashboard depends on a new API endpoint, the orchestrator waits for the endpoint task to complete before starting the dashboard.

# The dashboard depends on the API endpoint being ready
sf task create --title "Build analytics API endpoint" --plan "Sprint 14 Features"
sf task create --title "Build analytics dashboard" --plan "Sprint 14 Features" \
  --depends-on "Build analytics API endpoint"

Automatic merge and review

When an agent completes a feature, the Steward agent takes over — running tests, checking for conflicts, and merging the branch. You get a merge request with a summary of changes, test results, and any issues flagged during automated code review.

Real-time visibility into parallel agents

The Stoneforge web dashboard shows every agent’s progress in real time. See which features are in progress, which are in review, and which have merged — all from a single view. Whether you run multiple AI agents on a small project or across a large monorepo, parallel AI development with Stoneforge keeps your entire team’s work visible and coordinated.