Configuration

Key settings in VS Code (Settings → RavenKit). Change these when you need a different model, provider, or paths.

Chat agent (Corvus)

  • ravenkit.agent.provider — ollama (default) or anthropic or openai
  • ravenkit.agent.model — Model name, e.g. qwen2.5-coder:7b, llama3.2, claude-sonnet-4-20250514. Use auto to let RavenKit pick the best available Ollama model.
  • ravenkit.agent.ollamaUrl — Ollama API URL (default http://localhost:11434)
  • ravenkit.agent.anthropicApiKey — Your Anthropic API key when using Claude
  • ravenkit.agent.openaiApiKey / ravenkit.agent.openaiBaseUrl — For OpenAI or compatible endpoints
  • ravenkit.agent.maxTokens — Max tokens per response (default 4096)
  • ravenkit.agent.temperature — Model temperature 0–1 (default 0.7)

Paths

  • ravenkit.paths.root — Root directory for RavenKit data (default .ravenkit). Some setups use .features.
  • ravenkit.paths.features — Directory under root for feature files (default features)
  • ravenkit.paths.cache — Cache/index directory under root (default cache)

Corvus behavior

  • ravenkit.corvus.additionalContext — Extra context added to the system prompt (e.g. project guidelines)
  • ravenkit.corvus.includeProjectContext — Whether to include detected project context (framework, language) in the prompt (default true)

Stacksolo (sync)

When using Stacksolo for feature sync:

  • ravenkit.stacksolo.enabled — Enable Stacksolo (default true)
  • ravenkit.stacksolo.autoSync — Auto-sync on changes (default true)
  • ravenkit.stacksolo.syncInterval — Interval in ms (default 300000)
  • ravenkit.stacksolo.conflictResolution — ask, local, or remote

Get started · Commands · Troubleshooting