← Back to changelog

⚙️ Config v1.3.9

v1.3.9

  • Fixed summarization.trigger schema to match documented and runtime-supported types

    • Previously only accepted token_count (which silently never fired at runtime)
    • Now correctly accepts token_ratio, remaining_tokens, and messages_to_refine via a discriminated union
    • token_ratio value constrained to 0–1, other types require positive integers

  • Added thinkingDisplay parameter for Anthropic models

    • Controls whether reasoning content is returned in responses for Claude Opus 4.7+ (which omits thinking by default)
    • Options: "auto" (default), "summarized", "omitted"
    • Available on anthropic and bedrock (Anthropic models) endpoints

  • Added xhigh to Anthropic effort levels

    • effort field now accepts "", "low", "medium", "high", "xhigh", "max"

  • Added Claude Opus 4.7 to default model lists

    • claude-opus-4-7 for Anthropic
    • anthropic.claude-opus-4-7 for Bedrock

  • Added text/x-markdown MIME type alias

    • Maps to text/markdown for consistent file handling