Now with multi-agent support

The terminal that
thinks alongside you

Forge brings AI agents into your development environment — running commands, reading errors, and refactoring code while you stay in control.

Download for Mac See how it works →
Forge Terminal A terminal that understands context
Active
~/projects/api $ forge run deploy --prod
✓ Building docker image
→ forge/api:latest
→ Pushing to registry...
✓ Deployed to production
→ app.forge.dev/api
 
── Agent suggestion ──────────
AI Your health check at /api/health
returned 200 in 43ms.
No rollback needed.
 
~/projects/api $
Active Agents 3 running in parallel
Live
refactor-agent running
test-writer idle
docs-agent review
68% complete — 4 files remaining
Forge Drive Session storage and history
Synced
session_2025-05-06.log 2.1 MB
refactor-api-auth.patch saved
deploy-config.yaml modified
Trusted by engineers at Vercel Stripe Figma Linear Notion Shopify
AI-native

Agents that run
commands, not just suggest

Most AI coding tools give you text to copy. Forge agents actually run commands in your terminal, read the output, and adapt — without leaving your workflow.

⚡
Autonomous execution
Agents run shell commands, parse stdout, and react to errors in real time.
🔍
Context-aware suggestions
Every suggestion is grounded in your actual project structure and git history.
🔐
Model-agnostic
Plug in any model: Claude, GPT-4, Gemini, or a local LLM running on your machine.
forge — refactor-agent
auth/middleware.ts
const token = req.headers['x-auth-token']
if (!token) return res.status(401).send('Unauthorized')
const user = jwt.verify(token, process.env.SECRET)
const token = req.headers.authorization?.split(' ')[1]
if (!token) throw new AuthError('Missing bearer token')
const user = await verifyToken(token)
req.user = user
next()
AI · Refactor reasoning
Moved from synchronous to async token verification to prevent blocking the event loop under load. Extracted error handling into a typed AuthError class.
Drive

Every session,
perfectly archived

Forge captures every command, output, and agent interaction. Search across months of terminal history in milliseconds — and replay any session exactly as it happened.

Start for free View docs
forge drive — search
⌘ docker run production...
docker run -d --name api forge/api:1.4.2
May 6 · 14:32 · branch: main
docker run --rm forge/migrate:latest
May 3 · 09:14 · branch: release/1.4
docker run --env-file .env forge/worker
Apr 29 · 16:55 · branch: feature/queue
Themes

Your terminal,
your palette

Five curated themes — each with its own background tone, syntax colors, and personality.

$ forge run tests
Running 84 tests...
✓ All passed (1.2s)
Horizon
$ git push origin main
Counting objects...
→ remote: deployed
Matrix
$ forge agent run
Loading context...
Ready for input
Dusk
$ cargo build --release
Compiling forge v1.4
⚠ 2 warnings
Ember
$ ls -la src/
total 48
drwxr-xr-x api/
Warm Black
What engineers say

"Forge replaced iTerm for me in two days. The AI context is actually useful — it knows what I just ran and why it failed."

KL
Kai Larsen
Lead Platform Engineer · Vercel

"The drive feature is underrated. I found a command from 3 months ago in 2 seconds. Would have taken me 20 minutes in bash history."

SM
Sofia Muñoz
Senior SRE · Stripe

"Running 3 agents in parallel while I review PRs. It's like having a junior engineer who only does the boring parts — correctly."

TN
Tae-yang Nam
Backend Engineer · Linear

Ready to build
with agents?

Free for individual developers. Teams from $18/seat/month.

Download for Mac Linux & Windows →