Use code LAUNCH60 to get 60% off your first purchase
Native desktop · Local-first
One workspace for every coding agent.
Run Claude Code, Codex, Kiro, OpenCode and Hermes — plus terminals, browser previews, project memory and multi-agent workflows — from one native desktop app. Local-first. Pay once.
Every AI tool in a different tab. Nothing remembers. Nothing coordinates.
Context lives in tabs
Five terminals, five windows. Switch and you lose your place.
No shared memory
Each agent starts cold. You re-explain the same project endlessly.
Strictly one at a time
Agents wait in line instead of working the task in parallel.
→
One graph
Agentry runs them as a team.
A single visual graph where agents plan, build, and verify together — locally, with memory that sticks.
One command center
Persistent memory
Parallel execution
The real app. Not a mockup.
These are captures of Agentry running the Standard Build pipeline: a planner, a coordinator, three parallel builders on different runners, an aggregator, a reviewer and a verifier.
agentry · standard build pipeline
agentry · home
What you are looking at
Builders A, B and C run on Codex, Claude Code and Kiro at the same timeHome tracks open tasks, session notes and runner health per projectRun history keeps every graph run replayable, with logs per node
Drag, wire, and run agents as nodes. See the whole pipeline — who's planning, who's building, what's blocked — at a glance.
Planner→Coordinator→Builders ×3→Verify
Parallel Builder Waves
Fan a task out to multiple builders at once. Three agents working the same goal simultaneously — not waiting in a queue.
builder A
builder B
builder C
Verify & Auto-Retry
A Verifier checks output and loops the work back until it passes.
5 Runners + Fallback
Not installed? Agentry auto-falls back to the next available runner.
Chat · Build · Plan Rail
Switch modes without losing thread — one rail for talking, shipping, planning.
Multi-Workspace
Keep every project isolated with its own agents, memory, and history.
Memory + Memory Graph
Persistent, queryable context
Skills Marketplace
Drop-in reusable skills
Session Replay
Rewind any agent run
In-App Browser Preview
See changes live
Built-in Monaco IDE
Edit without leaving
Terminals + Task Board
Kanban + live shells
Chat. Plan. Build. One composer.
Use the same prompt box for quick answers, step-by-step planning, or full implementation — then switch modes without losing the thread.
Chat
Ask questions, explore ideas, get explanations. Conversational AI with your full model picker.
you → explain the auth flow agent → The auth uses JWT with…
Plan
Create a step-by-step plan before any code is written. Review, adjust, then execute.
1. Refactor auth module
2. Add integration tests
3. Update API docs
Build
Describe the task and let agents implement it. Simple one-agent or graph multi-agent workflows.
auth.ts modified
auth.test.ts created
docs/api.md pending
Simple workflow
One agent handles it
Graph workflow
Multi-agent visual flow
Eco / Full mode
Lighter or deeper runs
Reasoning effort
Low · Med · High · Max
context/
Project rules, architecture notes, and state that persists across every session.
tasks/
Active tasks, done tasks, and what still needs attention — tracked in plain Markdown.
handoffs/
Agent-to-agent notes so one agent's output feeds the next without re-explaining.
decisions/
Architecture decisions, workflow choices, and the reasoning behind them.
Agentry remembers the project, not just the chat.
Project context stored as plain Markdown files — rules, tasks, handoffs, decisions, and known bugs. Memory survives across sessions and can be opened outside the app.
You get the full Agentry product for 14 days — not a stripped-down free plan. Create an account, start the trial from Monthly checkout (INR), then download the app and sign in. Cancel before day 14 and you won’t be charged. Creating an account alone doesn’t start the trial.
Is Agentry really a one-time purchase?
Yes. The Lifetime plan is pay-once and yours forever, including all future updates. Monthly and Yearly exist if you prefer to spread the cost, but most daily drivers pick Lifetime.
Does my code or data ever leave my machine?
No. Agentry is local-first with no cloud and no telemetry. Your code, API keys, and memory stay on your computer. The only network calls are the ones your chosen AI runners make.
Which agents and runners does it support?
Claude Code, OpenAI Codex, Kiro, OpenCode, and Hermes out of the box, plus Groq and NVIDIA API keys for extra chat models. If a runner isn’t installed, Agentry auto-falls back to the next available one.
What platforms does it run on?
macOS and Windows. The installer detects your platform and the download page highlights the right build automatically.
Do I need separate subscriptions for the AI models?
Agentry orchestrates the agents and runners you already have. You bring your own model access (e.g. Claude Pro, an OpenAI key); Agentry coordinates them in one graph.
What’s your refund policy?
A 7-day money-back guarantee. If Agentry isn’t for you, email us within seven days for a full refund — no hoops.
Claude Code
Codex
Kiro
OpenCode
Hermes
Groq · NVIDIA
