| name | ollama-cli-ask |
| description | Use the local 'ask' CLI to interact with Ollama models. Supports one-shot queries, interactive chat, piping context, and JSON output. |
Ollama CLI (ask) Skill 🦑
This skill leverages the ask CLI tool to interact with local Ollama models. It's faster and more flexible than raw curl requests.
Features
- Context-aware: Remembers conversation history in interactive mode.
- Piping: Can ingest file content or command output via stdin.
- Thinking: Supports
<think> tag parsing for reasoning models (DeepSeek, etc).
- JSON: reliable JSON output for tool use.
Installation (if missing)
cp ~/clawd/tools/ollama-cli/ask.sh ~/bin/ask
chmod +x ~/bin/ask
Recommended Models
- Coding:
qwen3-coder-next (Smartest local coder)
- General:
llama3.1:8b (Fast, good generalist)
- Reasoning:
deepseek-r1 (if available, use -t flag)
Examples
Summarize a file:
cat MEMORY.md | ask "Extract the key user preferences"
Generate JSON data:
ask --json "List 5 sci-fi book titles and authors"
Complex Reasoning:
ask -t "Solve this logic puzzle: Three gods A, B, and C are called..."