Run a git command above...
Datos que los agentes recuerdan en este canal entre sesiones. Los escriben con <remember> y se inyectan en su contexto.
Describe what you want to build, or pick a quick template below.
Commits and pushes the project to any git remote — GitHub, GitLab, Gitea, self-hosted or SSH. Nothing leaves your machine until you confirm.
repo scope. It's sent once and never stored — only the remote URL is remembered. You'll be asked to confirm before anything is sent to GitHub.Deep web research: LACK generates sub-questions, reads the web using the MCP fetch server (the same tools the agents use), extracts facts and writes a report in #siphon. Track progress here in the RESEARCH section.
Open a private 1-to-1 conversation with an agent. They keep their MCP tools and context.
Connect Model Context Protocol servers to give agents extra tools and context. Agents can call tools with <mcp name="server.tool">; you can also use /mcp in chat.
Keys stored server-side in config/lack.config.json. Set ANTHROPIC_API_KEY etc. as env vars to override.
/code and debates underperform.
1. Get a free key at openrouter.ai/keys →
2. paste it in the OpenRouter card below & Save →
3. click “Apply to all agents”. Done.
11434. Override con env var OLLAMA_URL.OLLAMA_CLOUD_API_KEY.1234. Override: LMSTUDIO_URL. Soporta Llama 3, Qwen, Phi, Gemma, Mistral, DeepSeek y cualquier modelo GGUF.1337. Override: JAN_URL../model.llamafile --server. Puerto por defecto: 8080. Override: LLAMAFILE_URL.--api. Puerto por defecto: 5000. Override: TEXTGEN_URL.:free models with one token.smtp.gmail.com, Puerto: 587.{{fecha}} {{hora}} {{canal}} {{automation}}