Frequently Asked Questions
General
What is Commander?
Commander is a multi-agent orchestration engine that coordinates multiple AI agents across 5 canonical topologies (single, chain, dispatch, orchestrator, review), backed by 25 LLM providers and 18 built-in tools.
How is Commander different from other AI coding tools?
Most AI coding tools use a single agent with a single model. Commander orchestrates multiple agents across any topology, automatically selecting the best execution strategy based on task complexity. It's production-oriented from day one — circuit breakers, multi-tenancy, crash-safe checkpoints, Prometheus metrics, and live SSE streaming.
Is Commander open source?
Yes. Commander is MIT licensed.
What does "multi-agent orchestration" mean?
Commander can spawn multiple AI agents that work together on a single task. Depending on the topology, they might work in parallel, review each other's answers, or follow a lead architect who delegates to specialists.
Setup
Do I need multiple API keys?
No. One key is enough. Set OPENAI_API_KEY (or Anthropic, DeepSeek, Groq, …) and Commander auto-detects the provider. Failover chains kick in when you configure multiple keys.
Can I run Commander locally without internet?
Yes, with local models via Ollama or vLLM:
export OLLAMA_BASE_URL=http://localhost:11434
npx tsx packages/core/src/cliEntry.ts run "analyze this code"Is the npm package available?
@commander/core and @commander/sdk are built for publish; public npm release is in progress. Until then, clone the monorepo and use workspace packages. See Agent SDK.
Usage
Can Commander edit my files?
By default yes — control it with approval modes:
plan— show plan onlyread-only— read files, no editssuggest— suggest changes, wait for approvalauto-edit— edit automaticallyfull-auto— fully autonomous
Can I run Commander in CI/CD?
Yes. Use full-auto mode:
export COMMANDER_MODE=full-auto
npx tsx packages/core/src/cliEntry.ts run "fix all lint errors"How many agents does Commander use?
1–20 agents depending on task complexity. Simple tasks get 1 agent; complex research or engineering tasks get a full team.
What are the five topologies?
| Topology | Use when |
|---|---|
| SINGLE | Simple, one-shot answers |
| CHAIN | Sequential pipeline steps |
| DISPATCH | Parallel specialists + synthesis |
| ORCHESTRATOR | Lead agent delegates to workers |
| REVIEW | Produce then critique / merge |
Performance
Is Commander fast?
Simple tasks often complete in seconds. Complex multi-agent runs typically take tens of seconds to minutes depending on models and tools. Semantic caching (SHA-256 + similarity) reduces redundant LLM work.
Enterprise
Does Commander support multi-tenancy?
Yes. Per-tenant rate limits, concurrency quotas, storage isolation, memory sandboxing, and API key mapping. See Multi-Tenancy.
Can I deploy Commander on-premise?
Yes. Docker-based deployment works on any Linux VM. See Deployment.
Is there a paid tier?
The open-source version is fully featured and free. Managed cloud (SSO, hosted API, SLA) is on the roadmap.
Data & Privacy
Does Commander store my code?
Commander runs on your machine (or your server). It does not upload code to a Commander cloud by default. Content is only sent to the LLM providers you configure (unless you use fully local models).
Can I use Commander with sensitive code?
Yes. Prefer Ollama/vLLM or a VPC deployment so no code leaves your environment.