moorcheh-edge answer

Synopsis

moorcheh-edge answer --query TEXT [options]

Embeds --query locally (BGE-small-en-v1.5, 384-dim), searches the store, and calls POST /answer. The server uses Ollama with fixed model qwen2.5:0.5b-instruct. When search returns no passages (empty store or --kiosk-mode + --threshold filters everything), the server skips the LLM and returns: I don't have enough information to answer that question.

Run moorcheh-edge up first so Ollama is installed and the LLM model is pulled. Use --skip-ollama on up only if you want search without RAG.

Options

Flag	Default	Description
`--query`	Required	Question to answer
`--top-k`	`5`	Passages to retrieve for context
`--threshold`	`0.0`	Minimum score when `--kiosk-mode` is set
`--kiosk-mode`	off	Filter low-scoring passages
`--header-prompt`	-	Custom system instruction
`--footer-prompt`	-	Instruction before the question
`--chat-history-json`	-	Prior turns as JSON array
`--temperature`	`0.2` (server)	LLM temperature 0.0–2.0
`--timeout`	`120`	HTTP timeout in seconds
`--base-url`	`http://localhost:8080`	API base URL

Examples

moorcheh-edge upload-documents --documents-file documents.json
moorcheh-edge answer --query "Who won the football match?" --top-k 5 --kiosk-mode --threshold 0.8

Off-catalog question (no matching chunks):

moorcheh-edge answer --query "What is the weather on Mars?" --kiosk-mode --threshold 0.8

With custom temperature and longer timeout (useful on edge hardware):

moorcheh-edge answer --query "Who won?" --top-k 5 --temperature 0.1 --timeout 300

Output

Prints JSON to stdout:

{
  "answer": "...",
  "model": "qwen2.5:0.5b-instruct",
  "query": "Who won the football match?",
  "context_count": 1,
  "sources": [...]
}

See API: Answer.

Getting Started

Runtime

Search

AI Generation

Data Operations

Delete

Voice

Synopsis

Options

Examples

Output

​Synopsis

​Options

​Examples

​Output

Synopsis

Options

Examples

Output