Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 20 additions & 24 deletions docs/install-mac.md
Original file line number Diff line number Diff line change
Expand Up @@ -89,10 +89,10 @@ codex
npm install -g quadwork@latest
```

Verify:
Verify the install:
```bash
quadwork --version
# You should see the version number (e.g., 1.14.5)
npm list -g quadwork
# You should see the installed version, e.g. quadwork@2.7.0
```

### Troubleshooting: `EACCES: permission denied`
Expand Down Expand Up @@ -136,45 +136,41 @@ quadwork init
quadwork start
```

You should see output like:
```
QuadWork dashboard: http://localhost:8400
```

Open the dashboard URL in your browser to access the web UI.
The dashboard starts on `http://127.0.0.1:8400` (the server binds loopback
only). QuadWork opens that URL in your default browser automatically; if it
doesn't, open it yourself to reach the web UI.

---

## Create Your First Project

1. Open the dashboard at `http://localhost:8400`
1. Open the dashboard at `http://127.0.0.1:8400`
2. Click **"+ New Project"** or navigate to `/setup`
3. Fill in the project details:
- **Name:** Your project name
- **Repo:** GitHub repo in `owner/repo` format
- **Working directory:** Absolute path to the repo clone
- **Agent backends:** Choose Claude, Codex, or Gemini for each agent role
- **Agent backends & models:** In the **Agent Models** step, pick a CLI
backend — **Claude Code**, **Codex**, or **Gemini CLI** — for each of the
four roles (Head, Dev, RE1, RE2), and optionally select a specific model
per role (e.g. `opus` / `sonnet` for Claude, `gpt-5.4` / `gpt-5.6-*` for
Codex, `gemini-2.5-pro` / `gemini-2.5-flash` for Gemini). Leave a role on
**(CLI default)** to use the backend's own default model.
4. Click **Create**

QuadWork will:
- Create worktree directories for each agent (e.g., `project-head/`, `project-dev/`, `project-re1/`, `project-re2/`)
- Seed AGENTS.md files for each role
- Create a git worktree for each agent next to your repo (e.g., `project-head/`, `project-dev/`, `project-re1/`, `project-re2/`)
- Seed `AGENTS.md` and `CLAUDE.md` into each role's worktree

---

## Trust Prompt (Claude Code)

On first launch, Claude Code agents may get stuck at a "Do you trust this directory?" prompt. QuadWork v1.14.5+ automatically pre-trusts worktree directories for Claude-configured agents during project creation.

**If agents are still stuck** (e.g., upgraded from an older version), manually pre-trust each worktree:

```bash
# Run in each worktree directory
cd /path/to/project-head && claude -p "echo ok"
cd /path/to/project-dev && claude -p "echo ok"
cd /path/to/project-re1 && claude -p "echo ok"
cd /path/to/project-re2 && claude -p "echo ok"
```
On first launch, Claude Code shows a "Do you trust this directory?" prompt.
QuadWork auto-answers it: when an agent's terminal starts, a listener watches
its output for the trust prompt (within the first few seconds) and confirms it
automatically, so Claude-backed agents reach a trusted session with no operator
action. No manual pre-trust step is required.

---

Expand Down
Loading