Configuration Guide

DevDiff supports flexible configuration using a .devdiff.config.js file (recommended), .devdiff.config.json, or other .devdiffrc configurations in the root directory of your repository.

---

Configuration Schema

Here is a full schema reference detailing all available options in .devdiff.config.js format:

// .devdiff.config.js
export default {
  ai: {
    providers: [
      {
        name: "ollama-local",
        url: "ollama://llama3.2:3b",
        priority: 1,
      },
      {
        name: "openai-cloud",
        url: "openai://gpt-4o-mini",
        apiKey: process.env.OPENAI_API_KEY,
        priority: 2,
      },
    ],
    routing: {
      strategy: "priority",
      complexityThreshold: 0.6,
      localOnly: false,
    },
  },
  exclude: [
    "**/node_modules/**",
    "pnpm-lock.yaml",
    "package-lock.json",
    "dist/**",
    "*.log",
  ],
  cache: {
    enabled: true,
    path: ".devdiff/cache.json",
  },
  format: "markdown",
};

---

Configuration Options

Option	Type	Default	Description
ai.providers	Array	Required	A list of model provider endpoints and credentials. Supports local (ollama://) and cloud (openai://, gemini://, anthropic://).
ai.routing.strategy	string	"priority"	Strategy used by the Intelligent Router. Options: 'priority', 'cost', 'latency'.
ai.routing.complexityThreshold	number	0.6	Complexity value above which changes trigger escalation to higher capacity models.
ai.routing.localOnly	boolean	true	When true, prevents the router from selecting any external cloud providers regardless of complexity.
exclude	Array<string>	[]	Glob patterns matching directories, lockfiles, or assets to ignore during git diff scans.
cache.enabled	boolean	true	When true, caches generated changelog outputs locally to prevent redundant LLM token costs.
cache.path	string	".devdiff/cache.json"	Path where the local cache file is stored.
format	string	"markdown"	Default output format for changelogs. Options: 'markdown', 'json', 'html'.

---

Multiple Provider Failover (High Availability)

The DevDiff router supports failover chains. If a local model fails to respond or a cloud provider hits rate limits, the router automatically attempts the fallback chain:

"providers": [
  {
    "name": "ollama-primary",
    "url": "ollama://llama3.2:3b",
    "priority": 1
  },
  {
    "name": "openai-fallback",
    "url": "openai://gpt-4o-mini",
    "priority": 2
  }
]

Under this configuration:

1. DevDiff initially routes requests to ollama-primary.
2. If the local Ollama service is offline or errors out, DevDiff immediately falls back to openai-fallback to complete the request seamlessly.