llms.txt
# 1DevTool — AI Context File
> 1DevTool is an AI-native desktop developer workspace (macOS, Windows, Linux) that consolidates 9 developer tools into one window: IDE, terminal, embedded browser, HTTP client, database client, Docker manager, Git client, file explorer, and markdown preview. It is purpose-built to run AI coding agents (Claude Code, Codex CLI, Gemini CLI, Amp, OpenCode, Cline, Qoder, Aider, or any shell command) with full project context, so your agent sees console errors, network requests, database rows, and screenshots without copy-pasting.
## Product Overview
1DevTool is a "bring your own AI" workspace IDE. It does not host AI or charge for AI usage — you install the AI CLIs you already pay for and 1DevTool runs them side-by-side with the rest of your toolchain in a single window. Sessions survive restarts via tmux-backed terminal persistence, and every tool in the app (browser, HTTP client, DB client, etc.) has a one-click "Send to AI" that streams context directly into your agent's input.
**Tagline:** "9 dev tools. One window. Your AI sees everything."
**Positioning:** Alternative to Cursor, GitHub Copilot, Windsurf, Zed, JetBrains AI. Unlike subscription-based AI IDEs, 1DevTool is a **one-time $29 license** — no subscription, no per-seat pricing, no AI token upcharge.
**Supported platforms:**
- **macOS** 11+ (Apple Silicon M1/M2/M3/M4 — recommended — and Intel)
- **Windows** 10/11 (64-bit — NSIS installer + portable `.exe`)
- **Linux** (Debian/Ubuntu `.deb` + universal AppImage)
**System notes:** Full PTY terminal support on all platforms. Tmux-based session persistence is available on macOS and Linux only.
**Works with any AI agent** that runs as a CLI: Claude Code, Codex, Gemini CLI, Amp, OpenCode, Cline, Qoder, Aider, plus bash / zsh / PowerShell, plus any custom command you define.
---
## Core Concepts
### Projects (Workspaces)
A "project" in 1DevTool is a folder on disk with associated state: file tree, named terminals, browser tabs, HTTP collections, saved DB connections, environment files, and git config. Projects are color-coded and switch with a single keystroke. Everything inside a project — including running AI agent sessions — persists across app restarts.
### Terminals (AI Agent Panes)
Every terminal is a full PTY that can run any command: a shell, a Claude Code session, a dev server, a test runner, or a custom CLI. Terminals can be tiled into grids, columns, or fullscreen. On macOS and Linux, terminals are backed by tmux so sessions survive app restarts, crashes, and reboots.
### Channels
Channels are Discord-style AI conversation threads inside a project. Each channel holds an ongoing back-and-forth with one or more agents. Channels can be paused, resumed, templated, and piped into each other.
### Send to AI
A one-click action available in the browser, HTTP client, database client, and screenshot annotator. It packages up the current context (URL, console logs, network requests, SQL rows, annotated image, etc.) and drops it into any terminal's agent input — across every project in your workspace.
### Skills
Reusable AI agent instructions (prompt packs, tool configs, workflows). 1DevTool ships a Skills Manager for browsing, installing, editing, and creating skills that any agent can use.
### Agent Pipelines
Chains of AI agents across terminals with conditional branching (regex/contains match on output), delay/checkpoint nodes, output-capture modes, and transform steps between nodes. Lets you compose multi-agent workflows (e.g., planner → implementer → reviewer) without leaving the app.
### Templates
Pre-built project starters that install via the `onedevtool://` deep-link scheme. Categories include landing pages, blogs, docs, directories, status/changelog sites, SaaS, and desktop apps.
---
## Features
1DevTool ships 86 features across 12 categories. The sections below document every feature area with usage instructions.
### 1. Terminal & AI Agents
**Run any AI CLI or shell in a first-class pane.**
**Supported agents out of the box:** Claude Code, Codex, Gemini CLI, Amp, OpenCode, Cline, Qoder, Aider. Any other CLI (including local models via Ollama, LM Studio, or llama.cpp) can be added as a **custom agent** by specifying the command line.
**Opening a terminal:**
1. Open a project.
2. Click the "+" in the terminal tab bar, or press the terminal shortcut.
3. Pick an agent from the launcher (Claude Code, Codex, shell, custom), or type a command.
**Terminal features:**
- **Multi-agent terminals** — Run Claude, Codex, Gemini, Amp, and shells side-by-side in the same project.
- **Grid / column / fullscreen layouts** — Tile terminals or pop one into focus.
- **Terminal Dashboard** — Bird's-eye Kanban view of every running terminal across every project. See who's waiting on input, who's running, who's done.
- **Session persistence (tmux)** — On macOS/Linux, terminals survive app restarts, crashes, and OS reboots. Reopen the app, open the project, sessions are still there with full scrollback.
- **Session continuity** — Resume a past AI session from any agent. Resume a Claude Code transcript inside Codex, or a Gemini transcript in Amp, without re-explaining context.
- **Activity Logs** — Cross-project feed of terminal completions and AI-generated files. Click an entry to jump back into the terminal that produced it.
- **Channel Chat** — Discord-style channel view of AI conversations with pause/resume.
- **Channel Templates** — Save and share channel prompt configurations.
- **Drag-to-Agent-Input** — Drag files, screenshots, or URLs onto a terminal's agent input overlay.
- **Send File to Terminal** — Pipe a file's path into the active agent with one click.
- **Native OS notifications** — Get notified when an agent finishes or requires input, even in the background.
- **Remote Terminal** — Control any terminal from a paired phone (see Remote Control below).
- **Terminal Reader Mode** — Distraction-free scrollback reader with wrapped long output.
- **Sticky Notes Reader Mode** — Pin important lines from terminal output as sticky notes.
- **Hide Terminal** — Collapse a terminal without killing its tmux session.
- **Terminal Record** — Record a terminal session and export as MP4 video for sharing.
### 2. Project Management & Workspace
**Manage many projects without losing context.**
- **Multi-project workspace** — Open multiple projects simultaneously. Each gets its own color, layout, terminals, browsers, and connections.
- **Smart file explorer** — File tree with git status overlay, search, and drag-to-agent support.
- **Code intelligence (LSP)** — Go-to-definition, find references, hover docs across 10 languages. Cmd+Click works across files and libraries.
- **Color-coded projects** — Each project has a color tag shown in the sidebar and window chrome.
- **Layout presets** — Save terminal + panel arrangements and switch between them.
- **Remote Control** — Pair your phone via QR code. Send prompts, monitor agents, and trigger commands from anywhere.
- **Mobile Remote Access** — View and control running AI sessions from your phone's browser.
- **Built-in deploys** — One-click Vercel and Cloudflare Pages deploys from the project sidebar.
- **SSH Manager** — Save, organize, and connect to SSH hosts from a managed connection list.
- **Remote SSH Projects** — Open a remote folder as a project over SSH.
- **Clone from GitHub account** — Linked GitHub accounts populate a repo picker; clone directly into a new project.
- **Git visual changes** — Visual diff + commit graph + blame.
- **Git worktrees** — Create and switch between worktrees with full env file copying.
- **Multi-account git** — Use different GitHub/GitLab identities per project.
### 3. HTTP Client
**Full Postman-style HTTP client inside your workspace.**
- **Request builder** — Method, URL, params, headers, body (JSON, form, multipart, binary).
- **Response inspector** — Status, headers, body with JSON/XML/HTML pretty-printing, timing breakdown.
- **Request workspace** — Tabs + nested collections, per-project or per-environment.
- **Import / export** — Reads Postman, Insomnia, and Bruno collection formats; exports back to the same.
- **Send to AI** — Drop a request + response pair into any terminal agent for debugging.
### 4. Database Client
**Visual database admin for 13 engines (SQL, NoSQL, cache, vector, and time-series).**
**Supported engines:** PostgreSQL, MySQL, MariaDB, SQL Server, CockroachDB, ClickHouse, TimescaleDB, Neon, Supabase, Amazon Redshift, MongoDB, Elasticsearch, CouchDB, Cassandra, Redis, Valkey, KeyDB, SurrealDB, Weaviate, InfluxDB, Kafka. (The marketing copy groups these as "26 engines" when counting variants and managed providers.)
- **Multi-database support** — Register connections for any supported engine. Credentials are stored in the OS keychain.
- **Query editor** — SQL editor with autocomplete, schema inspection, and result paging. NoSQL engines get engine-specific query editors.
- **Schema browser** — Tables, views, indexes, columns, foreign keys rendered as a tree.
- **Send to AI** — Send a failing query + schema + sample rows to an AI agent for debugging.
### 5. Docker Manager
**Inspect and control Docker without leaving the app.**
- **Containers** — Start, stop, restart, delete. Inspect env, ports, mounts, networks.
- **Logs & inspect** — Stream logs with search. Exec into a container.
- **Images & volumes** — List, prune, and inspect images/volumes.
### 6. Design Tool
**Lightweight design canvas for developers.**
- **Design canvas** — Sketch UI layouts directly in the app (no Figma needed for quick mockups).
- **AI design generation** — Generate initial layouts from a prompt; refine visually.
- **Design modes** — Switch between low-fidelity mockup and high-fidelity rendering.
- **Send design to AI** — Ship the canvas + notes to a coding agent to generate code.
### 7. Skills Manager
**Browse, install, edit, and create reusable AI agent skills.**
- **Skills browser** — Explore community and built-in skills.
- **Skills editor** — Author new skills with prompt, tools, and metadata.
### 8. Developer Toolbox
**11 offline utilities, available without opening the browser.**
- JSON Formatter & Validator
- JWT Decoder
- Regex Tester
- Diff Viewer
- Base64 Encoder/Decoder
- URL Encoder/Decoder
- Hash Generator (MD5, SHA-1, SHA-256, SHA-512)
- UUID Generator
- Timestamp Converter
- String Case Converter
- Markdown Preview
### 9. Browser Preview
**Embedded Chromium browser with developer capture hooks.**
- **Embedded browser** — Full page rendering with tabs.
- **DevTools access** — Open Chrome DevTools against any loaded page.
- **Send page to AI** — Capture screenshot (including lazy-loaded and JS state), annotate, and send with console logs + URL context to any terminal agent in one click.
### 10. Environment & Ports
- **Env Manager** — Edit `.env` files with syntax highlighting. Visual diff between environments. Copied into git worktrees automatically.
- **Port Manager** — See which port each dev server is running on. Kill a port. Detect port conflicts before running new services.
### 11. AI Productivity
Tools that wrap around whatever AI agents you run.
- **Prompt history** — Searchable history of every prompt you've typed into any agent.
- **Resume UI** — Visual picker for prior sessions; resume in the same or a different agent.
- **Combine sessions** — Merge context from multiple past sessions into one.
- **AI diff panel** — Review agent-generated code changes as a structured diff before applying.
- **Screenshot annotator** — Crop, highlight, and annotate screenshots; send to AI.
- **Command palette** — Cmd/Ctrl + K fuzzy-find projects, files, settings, and actions.
- **AI account switcher** — Log into multiple Claude / OpenAI / Google accounts; switch per terminal.
- **AI usage dashboard** — Track token usage and estimated cost per agent, per project, over time.
- **AI memory manager** — Browse, edit, and graph the cross-project memory your agents have built up.
### 12. Customization
- **Themes** — Light/dark + custom themes.
- **Keyboard shortcuts** — Fully rebindable.
- **Terminal fonts** — Font family, size, and ligature controls.
- **MCP settings** — Configure Model Context Protocol servers per agent.
---
## How to Use (Common Workflows)
### Quick Start
1. **Download** from https://1devtool.com/download. macOS Apple Silicon `.dmg` is recommended on Mac; Windows installer (`.exe`) on Windows; `.deb` on Ubuntu/Debian, AppImage elsewhere.
2. **Open a folder** — File → Open Project, or drag a folder into the sidebar.
3. **Launch an agent** — Click "+" in the terminal bar, pick Claude Code / Codex / Gemini / shell / custom.
4. **Open the browser** — `Ctrl/Cmd + 2` (or click the browser tab) and load a local dev server.
5. **Send context** — Click "Send to AI" on any browser page, HTTP response, or DB row to pipe it into your active terminal agent.
### Running Multiple AI Agents in Parallel
1. Open terminals side-by-side (grid or column layout).
2. Launch Claude Code in the left pane, Codex in the middle, a local shell in the right.
3. Use the Terminal Dashboard (`Cmd/Ctrl + D`) to monitor which agent is running vs waiting.
4. When one finishes, a native OS notification fires; Activity Logs surface the completion with a one-click jump-back.
### Debugging with "Send to AI"
1. Open a browser tab in 1DevTool and reproduce the bug.
2. Screenshot annotator auto-captures the viewport (including lazy-loaded content).
3. Annotate what's wrong (arrows, boxes, text).
4. Click **Send to AI** → pick target terminal.
5. The screenshot, URL, console logs, and network log are dropped into the agent's input; ask the agent to fix it.
### Resuming a Past Session
1. Open the Resume UI (`Cmd/Ctrl + R` in a terminal).
2. Pick any past session from any agent.
3. Optionally switch the target agent (e.g., resume a Claude transcript inside Codex).
4. The transcript is replayed into the new session as context.
### Deploying with Built-in Deploys
1. Sidebar → Deploys tab.
2. Click **Connect Vercel** or **Connect Cloudflare Pages** — OAuth device flow.
3. Select project and branch.
4. Click **Deploy** — status streams live; deploy logs are captured and available to Send to AI.
### Managing a Database Visually
1. Database client → "Add Connection."
2. Pick engine (Postgres, Mongo, Redis, etc.) and enter credentials.
3. Browse schemas, open a query editor, run a query, inspect results.
4. Click **Send to AI** on any query result to ask an agent to explain, rewrite, or debug.
### Remote Control From Your Phone
1. Sidebar → Remote Control → "Pair Device."
2. Scan the QR code with your phone's camera.
3. Open the resulting URL on your phone.
4. Send prompts, monitor terminals, and trigger commands from your phone. Works over LAN.
### Installing a Template
1. Visit https://1devtool.com/templates and pick a template.
2. Click "Install in 1DevTool" — the `onedevtool://` deep link opens the app and starts installation.
3. The template is cloned into a new project with dependencies installed.
---
## Licensing & Pricing
- **Free tier (permanent, no time limit):** Unlimited prompts to local AI terminals, Agent @mention orchestration, 1 project, 4 terminals, 1 browser tab, 1 git worktree, 1 database connection, 5 saved HTTP requests, 5 AI diffs per day, and 7-day prompt history, session resume, and AI memory. No credit card.
- **1 Device — $29 one-time** ($29/device)
- **3 Devices — $59 one-time** ($19.67/device, most popular)
- **5 Devices — $89 one-time** ($17.80/device)
**Every paid tier includes:** Unlimited projects, terminals, embedded browser tabs, git worktrees, database connections, saved HTTP requests, AI diffs, prompt history, session resume, and AI memory; Project Settings; project groups; custom layout, themes, and shortcuts; config export/import; deploy app; remote/mobile access; full v1 + all v1.x updates (1 year); priority support.
**Subscription?** No. One-time purchase. App keeps working forever after the 1-year update window; renewal is optional if you want new features past that window.
**BYO AI:** 1DevTool does **not** charge for AI usage. You bring your own API keys or subscriptions (Anthropic, OpenAI, Google, Ollama, etc.). The $29 license is for the desktop workspace itself.
**Checkout:** Lemon Squeezy (secure, 7-day money-back refund).
**Total cost example:** $29 one-time + your existing AI subscription (e.g., $20/mo Claude Pro) = significantly less than $20/mo Cursor + separate terminal + separate DB client.
---
## Download & Install
All builds are published at https://github.com/stoicsoft/1devtool-releases/releases and linked from https://1devtool.com/download.
| Platform | Recommended | Alternate |
|----------|-------------|-----------|
| macOS (Apple Silicon) | `.dmg` | `.zip` |
| macOS (Intel) | `.dmg` | `.zip` |
| Windows 10/11 (64-bit) | NSIS installer `.exe` | Portable `.exe` |
| Linux (Debian/Ubuntu) | `.deb` | AppImage |
| Linux (other distros) | AppImage | — |
macOS builds are notarized and hardened-runtime signed.
---
## Troubleshooting
### Terminals & AI Agents
**Agent command not found**
- Install the agent CLI globally first (`npm i -g @anthropic-ai/claude-code`, `codex`, `gemini-cli`, etc.).
- In 1DevTool, verify the command is on the PATH the app inherits — use the agent launcher's **Test Command** button.
- For custom agents, provide the absolute path to the binary.
**Session did not persist after restart (macOS / Linux)**
- Session persistence requires tmux to be installed and reachable. Run `which tmux` in an external terminal.
- If tmux is installed but sessions still lost, check Activity Logs → Session Continuity for the last snapshot; restore manually.
- Session persistence is **not available on Windows**. Windows terminals are killed on app close (by design, until a Windows-native persistence layer ships).
**Agent is stuck "waiting"**
- Terminal Dashboard shows each agent's state. A "waiting" agent is blocked on user input.
- Open the terminal and respond, or use Remote Control to respond from phone.
### Projects & Workspace
**File explorer shows stale files**
- Click the refresh icon in the file tree header, or `Cmd/Ctrl + R` inside the explorer panel.
- Large monorepos: check `.gitignore` — 1DevTool respects it to skip node_modules, .next, etc.
**Go-to-definition not working**
- LSP servers for the project's language must be installable. First time opening a project, 1DevTool downloads the LSP — check the bottom status bar for progress.
- Supported languages: TypeScript/JavaScript, Python, Go, Rust, Ruby, Java, C/C++, PHP, C#, Swift.
**Deploys fail with "Node version" errors (Vercel)**
- v1.15.1+ includes a Node version fallback. Update to the latest release.
- As a workaround, set `NODE_VERSION` in the deploy's env vars.
### HTTP Client & Database
**Import Postman collection produces errors**
- 1DevTool supports Postman v2.1 collection format. Older v1 formats must be exported as v2.1 first.
- Pre-request scripts and test scripts are **not executed** (inspection-only).
**DB connection times out**
- Check the engine-specific `connectTimeoutMS` / `connect_timeout` setting.
- For cloud databases (Neon, Supabase, Redshift), verify your IP is allowlisted.
- SSH tunneling: use SSH Manager to open a tunnel, then point the DB connection at `localhost`.
### Browser & Send-to-AI
**Send to AI produces empty context**
- The browser panel must be the active source. Click the browser before clicking Send to AI.
- For DB/HTTP panels, ensure a query/request has completed successfully — failed requests still get their error body sent.
**Screenshot capture misses content below the fold**
- Use **Full Page Screenshot** (the camera icon with ↓ arrow), which auto-scrolls before capturing.
### Licensing
**License key rejected**
- Verify you copied the full key (Lemon Squeezy keys are ~35 chars including dashes).
- Eac
