Agents
Navin agents: local AI team, parallel missions and @mentions
August 2, 2026 · 2 min read · Navin team
Complete guide to Navin agents: virtual teams, @mentions, model routing, multi-agent missions and local execution on your machine.
You want a local AI agent that truly works in your repo - not a chat inventing files. Navin orchestrates agents, virtual teams and parallel missions on Windows, macOS or Linux.
This is the reference guide for how Navin agents are organized, how @mentions work, how model routing saves cost, and how to run a multi-agent mission to a deliverable - without leaving the app.
Why Navin agents change the game
An in-editor assistant completes a line. A Navin agent reads the workspace, runs tools, opens the terminal, updates tasks and iterates until delivery. You can run several in parallel, each with a role, a model budget and a file scope.
The workflow becomes:
- Define a measurable goal (landing, audit, scrape, deck).
- Pick an agent or virtual team.
- Run locally with BYOK or optional managed keys.
- Collect results in chat, files, Vision 360 and Tasks.
Also read Vision 360 and the Code module.
Built-in specialists and @mentions
Navin ships domain agents. An @mention calls the right role with skills and slash commands:
- Forge for implementation
- Vector for SEO (marketing)
- Scribe for documents (documents)
- Scraping / research agents (scraping)
- Lead agents (leads)
Virtual teams on one mission
For a launch, spawn 2 to 5 roles as sub-agents. Deliverables merge at the end. That is a local AI team.
Mission anatomy
- Research agent (market, competitors)
- Forge (landing + instrumentation)
- Scribe (investor deck)
- Vector (SEO plan)
- Leads (ICP list)
Before building, many teams open RiskLens for a pre-mortem, then let agents execute a stronger plan.
Smart model routing
| Task type | Model bias |
|---|---|
| Pulse / summaries | Light / Flash |
| Feature build | Balanced executor |
| Architecture / security | Expert |
Agents + Code + studios
Agents plug into Code modes, Vision 360 and Leads / Marketing / Documents / scraping studios.
Comparison
| Need | Chat-only IDE | Navin |
|---|---|---|
| Multi-agent | Rare | Local parallel |
| AppSec | External | Security mode |
| Scraping & leads | Other tools | Built-in |
| Data | Often cloud | Local-first |
FAQ
Does Navin replace Cursor or Copilot?
It goes further on multi-agents, AppSec, scraping, documents and GTM in one local app.
Do agents upload my repo?
Workspace stays local. Model calls use your keys.
How many agents in parallel?
Depends on plan and machine. Start with 2-3 clear roles.
Do I need Discord?
No. Follow docs, blog, changelog on navin.live.
Implementation notes for agents-navin-equipe-ia-locale
Keep missions short. Prefer evidence over vibes. Use Plan before large edits, Review and Security before merge, and Vision 360 to watch preview plus tasks. Local-first means your repo stays on disk while model calls follow the keys you configure.
When you publish marketing or lead outreach, apply the same review discipline you would apply to code: facts, permissions, and a clear owner.
Try Navin on your machine
Local agent, cross-platform. Code, debug, scrape, leads, security and review - without leaving Navin.