Ollama
Run large language models locally.
Category AIPlatforms Linux / macOS / WindowsPrivileges Install needs rootChinese docs Many Chinese tutorialsStars ~182K
One-click install command
curl -fsSL https://ollama.com/install.sh | sh
Notes
After install, pick a model for your hardware, then run ollama pull / run.
Agent Prompt
Copy the block below and hand it to an AI Agent — replace [bracketed] parts with your own values.
Deploy Ollama (Run large language models locally) in my environment. Before starting: 1. Run read-only checks first: OS version, CPU architecture, available disk/RAM, port usage, Docker or sudo access — and report the findings. 2. Use the project source https://github.com/ollama/ollama as ground truth; verify the current stable version, prerequisites and compatibility. 3. Present an execution plan, the files/services/firewall rules to be changed, and a complete rollback plan; wait for my confirmation before executing. 4. Download remote scripts locally for review first; pin tag / image digest / commit; don't rely on latest or master. 5. Never put secrets (passwords, API keys, tokens, subscription URLs) in the prompt, shell history or logs — use 600-permission env files, Docker secrets or _FILE variables instead. Verified starting command (don't run blindly — download/expand and review first): curl -fsSL https://ollama.com/install.sh | sh Project-specific notes: - After install, pick a model for your hardware, then run ollama pull / run. - Permission requirement: Install needs root. Don't escalate privileges; don't run containers as root unnecessarily. After execution, verify and report back: - Installed version, service status and health check results; - Open ports, listen addresses, firewall changes; - Config files, data volumes, log and backup locations; - Access method (for sensitive credentials, only state where they're stored — never print them); - Restart policy, plus verified uninstall / rollback commands.