Skip to content

1.0.8

Latest

Choose a tag to compare

@github-actions github-actions released this 15 Nov 09:12
· 1 commit to main since this release
2e839b5

What's Changed

Added

  • Local models: --provider ollama --model <name> runs on Ollama at localhost:11434 (OLLAMA_BASE_URL to change it), and any other OpenAI-compatible server (LM Studio, llama.cpp, vLLM) works as --provider <url> --model <name>. No API key is needed, and --effort reaches Ollama as reasoning_effort. (#51)
  • Claude sign-in: Claude is a built-in provider, and Watchtower uses the Anthropic SDK's own credential lookup, so an API key (ANTHROPIC_API_KEY), a gateway token (ANTHROPIC_AUTH_TOKEN), an ant auth login profile, or workload identity federation all work. ANTHROPIC_MODEL_NAME sets the model (default claude-sonnet-5-5; --flagship uses claude-opus-5-5), and --provider claude is an alias for anthropic. (#52)

Fixed

  • Claude requests no longer send a temperature, which current Claude models reject; --provider anthropic failed on claude-sonnet-5-5. (#52)
  • --provider anthropic without --model uses a Claude model instead of gpt-5.6. (#52)
  • A custom URL provider no longer receives OPENAI_API_KEY when --apikey is not given; it gets a placeholder key. (#51)
  • --provider ollama works; it used to fall through to LiteLLM without the ollama/ prefix. (#51)

Changed

  • New logo: a watchtower with a laser eye, in the README, the docs header, and the favicon. (#50)
  • The documentation site moved to Fumadocs with a quiet shadcn look, at the same URLs, with built-in search. Its build fails on a broken internal link or anchor. (#53)
  • anthropic>=0.98.0 is now required, for ant auth login profiles and workload identity federation. (#52)

Full Changelog: 1.0.7...1.0.8