Agent Dispatch

CodeFactory dispatches local AI agents to fix code based on review feedback. Agents run entirely on your machine — your code never leaves your environment.

Supported Agents

CodeFactory automatically detects which CLI tools are available on your system and selects the most appropriate agent for each task.

How Dispatch Works

1. Worktree Isolation

Before an agent runs, CodeFactory creates a temporary git worktree:

/tmp/pr-babysitter/<repo>/<pr-number>/

This ensures:

• The agent works on an isolated copy of the branch.
• Your local working directory is never touched.
• Multiple agents can run in parallel on different PRs.

2. Agent Execution

The agent receives:

• The review feedback to address.
• The file context (relevant source files).
• The PR description for broader understanding.
• Any previous agent attempts (to avoid repeating failed approaches).

3. Validation

After the agent completes:

• Changes are diffed against the original branch.
• If configured, tests are run to validate the fix.
• The diff is logged for human review in the dashboard.

4. Commit & Push

If validation passes, CodeFactory:

• Creates a descriptive commit message explaining what was fixed.
• Pushes the commit to the PR branch on GitHub.
• Updates the feedback status to resolved.

Agent Runs in the Dashboard

Every agent run is tracked in the CodeFactory dashboard:

• Status — Running, succeeded, or failed.
• Duration — How long the agent took.
• Diff — Exact changes the agent made.
• Logs — Full agent output for debugging.

Model Discovery

CodeFactory discovers available models from your local CLI installations. You can view and refresh available models from the Settings page in the dashboard.

Customization

Override the default agent per repository or globally:

• Set CODEFACTORY_AGENT=claude or CODEFACTORY_AGENT=codex as an environment variable.
• Configure per-repo preferences in the dashboard under repository settings.

See Configuration for all options.