2.1Crew
The first agent leads
The first agent in your crew answers you. Midway through a task it brings in a crewmate for the part that crewmate does better.
Your agents, local models and API keys in one app, fully loaded: shared memory, scheduled flows, 55 playbooks, one click tools and a browser your agents can drive. It's free and runs on the plans you already pay for.
01The room
A room is one shared conversation for one task, with your agents in it. This is a real room from start to finish; click a line to jump there.
02How it works
2.1Crew
The first agent in your crew answers you. Midway through a task it brings in a crewmate for the part that crewmate does better.
2.2Briefings
A crewmate gets the goal, what the room has learned, the files that changed and its own task. Only the lead reads the whole log.
2.3Limits
When an agent runs out of usage, the next one takes the same task with a briefing. Lunchbox shows the reset time and brings the first one back after it.
2.4Proof
An agent can only say the tests pass when a test command passed after its last edit. Without one, the claim waits for you.
03Agents
OpenCode, Aider, Goose, Cursor agent, Qwen, Amp, Crush and Qoder.
Ollama, LM Studio and llama.cpp. Lunchbox can start Ollama and pull a model that fits your memory.
Anthropic, OpenAI, xAI, Google, OpenRouter, DeepSeek, Groq, Mistral, Together, Fireworks, Cerebras, Kimi, DeepInfra, Perplexity, NVIDIA and GitHub Models.
04Pool
Lunchbox reads each plan's real limits through that tool's own login. You see what is left in every 5 hour window and every week, when each one resets, and which agent did best at each kind of work.
05Savings
Lunchbox sets up every room for prompt caching, so tokens the provider has already seen cost about a tenth of the price. On a plan that means more room before your limit. On an API key it means a smaller bill.
The Savings page shows what you saved, per agent and per room.
06Side by side
One message goes to every agent in the room, and the answers arrive in columns. The one you keep feeds that agent's rating, so Pool shows which agent wins at which kind of work.
You
07Facts
.jsonl files on your disk. API keys sit in the system keychain.08Install