Built-in Module

ai.generate

Generate text using AI models. Supports Ollama, OpenAI, and custom endpoints.

Configuration

Parameter	Type	Required	Description
prompt	string	Yes	The prompt to send to the AI model
model	string	No	Model name (default: from config.yaml)
system_prompt	string	No	System instructions for the model
max_tokens	number	No	Max tokens in response

Provider Configuration

Configure your AI provider in .shiro/config.yaml:

Ollama (Local)

yaml

ai:
  provider: ollama
  model: llama3.2
  endpoint: http://localhost:11434

OpenAI

yaml

ai:
  provider: openai
  model: gpt-4o
  api_key: "{{env.OPENAI_API_KEY}}"

Custom Endpoint

yaml

ai:
  provider: custom
  model: my-model
  endpoint: https://my-ai-api.example.com
  api_key: "{{env.CUSTOM_API_KEY}}"

Examples

Code Review

json

{
  "id": "review",
  "type": "ai.generate",
  "config": {
    "system_prompt": "You are an expert code reviewer. Be concise and actionable.",
    "prompt": "Review this diff and identify issues:\n{{steps.get_diff.diff}}",
    "model": "gpt-4o"
  },
  "depends_on": ["get_diff"]
}

Commit Message Generation

json

{
  "id": "gen_commit_msg",
  "type": "ai.generate",
  "config": {
    "prompt": "Write a conventional commit message for:\n{{steps.get_diff.diff}}",
    "max_tokens": 100
  },
  "depends_on": ["get_diff"]
}

Output

text - Generated text response
model - Model used
tokens_used - Token count (if available)