Getting Started
This guide walks you from install to your first agent.
Prerequisites
- macOS (PurePoint is macOS-only for now)
- git with worktree support
- At least one AI coding CLI: Claude Code, Codex, or OpenCode
Install
Download the latest .dmg from Releases.
Open the DMG, drag PurePoint to Applications, and launch. The app installs the pu CLI to ~/.pu/bin/pu on first launch.
Add to your shell config (~/.zshrc or ~/.bashrc):
export PATH="$HOME/.pu/bin:$PATH"
Reload your shell, then verify:
pu health
Initialize a workspace
Navigate to any git project and initialize:
cd ~/my-project
pu init
This creates .pu/ with a manifest and default config. See Directory Structure for details.
Spawn your first agent
pu spawn "fix the typo in README" --name fix-typo
This:
- Creates a git worktree at
.pu/worktrees/wt-{id}on branchpu/fix-typo - Spawns a Claude agent in that worktree
- Sends the prompt to the agent
Check status:
pu status
Watch the agent's output:
pu logs ag-{id}
Or attach to the agent's terminal interactively:
pu attach ag-{id}
Press Ctrl+C to detach without killing the agent.
Use the macOS app
Open PurePoint from Applications. Add your project directory. You'll see:
- Sidebar: All worktrees and agents with live status indicators
- Terminal: Click any agent to view its terminal
- Command palette (
Cmd+N): Spawn new agents or swarms
Spawn with different agent types
pu spawn "refactor tests" --agent codex --name refactor # Codex
pu spawn "review the PR" --agent opencode --name review # OpenCode
pu spawn --root --agent terminal # Plain terminal
Spawn from a template
pu spawn --template code-review --var BRANCH=main --name review
See Templates and Definitions for creating templates.
Review agent work
View diffs across all worktrees:
pu diff
pu diff --worktree wt-abc12345 # specific worktree
pu diff --stat # file summary only
Clean up
Kill a specific agent:
pu kill --agent ag-xyz98765
Remove a worktree, its agents, and branches:
pu clean --worktree wt-abc12345
Clean everything:
pu clean --all
Next steps
- CLI Reference -- all commands and flags
- Configuration -- customize agent behavior
- Templates and Definitions -- reusable prompts and agent types
- Scheduling and Triggers -- automate agent workflows
- Concepts -- the mental model