vrk prompt

About

Sends a prompt to an LLM and prints the response to stdout. Works with Anthropic and OpenAI APIs. You pipe content in, optionally set a system prompt, and get the response back - no curl commands, no JSON escaping, no response parsing. Temperature defaults to 0 for deterministic output.

The problem

You need to call an LLM from a shell script. You write a curl command with JSON escaping, header management, and response parsing. The prompt changes and you break the JSON. The API returns an error and you parse it wrong.

Before and after

Before

curl -s https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "content-type: application/json" \
  -d '{"model":"claude-sonnet-4-6","max_tokens":1024,"messages":[{"role":"user","content":"Summarize this"}]}'

After

cat article.md | vrk prompt --system 'Summarize this' --model claude-sonnet-4-6

Example

cat article.md | vrk prompt --system 'Summarize this' --model claude-sonnet-4-6 --json

Exit codes

Code	Meaning
0	Success
1	API failure, budget exceeded, or schema mismatch
2	Usage error - no input, missing flags

Flags

Flag	Short	Type	Description
`--model`	-m	string	LLM model (default from VRK_DEFAULT_MODEL or claude-sonnet-4-6)
`--system`		string	System prompt text, or @file.txt to read from file
`--budget`		int	Exit 1 if prompt exceeds N tokens
`--fail`	-f	bool	Fail on non-2xx API response or schema mismatch
`--json`	-j	bool	Emit response as JSON envelope with metadata
`--quiet`	-q	bool	Suppress stderr output
`--schema`	-s	string	JSON schema for response validation
`--explain`		bool	Print equivalent curl command, no API call
`--retry`		int	Retry N times on schema mismatch (escalates temperature)
`--endpoint`		string	OpenAI-compatible API base URL