Kilo-Org/kilocode v5.8.0

Release v5.8.0

on GitHub

5.8.0

Minor Changes

• #5247 12ea08e Thanks @theQuert! - Add Apertis as a new API provider

• #5883 31b271e Thanks @Astricaelus! - Add MiniMax-M2.5, MiniMax-M2.5-highspeed and MiniMax-M2.1-highspeed models

• #5526 fe86f25 Thanks @nubol23! - Added Voyage AI embedder support

Patch Changes

• #5878 74a9d8d Thanks @CoinAnole! - Fix Moonshot coding endpoint model selection so it includes all Moonshot models while keeping kimi-for-coding hidden on non-coding endpoints.

• #5746 38ea95b Thanks @Githubguy132010! - Add dev container persistence for threads and settings

• #5920 4a11f4f Thanks @sebastiand-cerebras! - Remove deprecated Cerebras models: llama-3.3-70b and qwen-3-32b

• #5575 c43b9b4 Thanks @Patel230! - fix: treat maxReadFileLine=0 as unlimited (same as -1)

• #5639 853b03f Thanks @mikij! - Updates some visual bugs in Agent Behaviour settings page

• #5195 d463a8b Thanks @Drilmo! - Filter internal verification tags from assistant messages before displaying to users

• #5904 25da368 Thanks @Githubguy132010! - Fix scroll jump issue when reading long completion messages in Agent Manager

• #5811 90a34f6 Thanks @wombatepiclandingstudio! - fix: prevent context token indicator flickering

• #5565 d58ba04 Thanks @Githubguy132010! - Fix file deletion auto-approve checkbox not being clickable

• #5377 cf9d90e Thanks @benzntech! - Fix recurring MODEL_NO_TOOLS_USED error loop by detecting text-based tool call hallucinations and instructing the model to use the native API.

• #5586 ffb7d87 Thanks @abdulrahimpds! - Fixed UI issues in Settings search bar: clipping of results and layout shift when expanding

• #5760 ebbb072 Thanks @Githubguy132010! - Fix user message visibility by using distinctive theme-aware colors

• #5752 3db4b15 Thanks @Madrawn! - fix(mentions): process slash commands in tool_result blocks

• #5055 74d904e Thanks @Leoyzen! - feat: support preserving reasoning content in OpenAI format conversion

• #5739 fc02342 Thanks @rayss868! - Fix: Honor explicit 'disable' for reasoning effort

• #5167 8b9de69 Thanks @hdcodedev! - Fix: "Kill Command" button now reliably terminates processes on all platforms, including those running in the background.

• #5953 45a5414 Thanks @kevinvandijk! - Add support for Claude Sonnet 4.6 to Anthropic, Bedrock, and Vertex providers (thanks @PeterDaveHello & Roo)

• #5630 29f287a Thanks @SenatusSPQR1! - fix(nano-gpt): Add native reasoning field extraction

• #5814 5b10436 Thanks @shssoichiro! - Support custom embed dimensions for Ollama provider

• #5523 5610aca Thanks @abdulrahimpds! - Fix: Persist total API cost after message deletion

• #5926 1ca21c8 Thanks @alex-alecu! - Review mode now offers one-click suggestions to apply fixes, switching to Code, Debug, Architect, or Orchestrator mode based on the review findings

• #5897 0a3fae1 Thanks @evanjacobson! - fix: prevent MCP servers from restarting on every settings file re-read

• #5150 101b672 Thanks @hdcodedev! - fix: prevent duplicate tool_use/tool_result IDs in conversation history (#4482)

• #5896 d7fc4cb Thanks @evanjacobson! - fixed falsy provider settings leak between profiles when switching

• #5149 d39b51c Thanks @dacsang97! - Enhance Anthropic extended thinking compatibility

• #5816 e33a2ef Thanks @evanjacobson! - Fix tool use failure for providers returning numeric tool call IDs (e.g. MiniMax) by coercing ID to string in the shared stream parser

• #5576 b67879e Thanks @Patel230! - fix: improve symlink handling in skills directory

• #5900 2f47a5f Thanks @evanjacobson! - prevent abort listener memory leak in attemptApiRequest

• #5459 4f2a7da Thanks @Schrolli91! - Implement better formatting for low cost values

• #5831 f8f8708 Thanks @Neonsy! - Fixed ZenMux context window detection to prevent erroneous context-condensing loops.

• #5831 26e4e1a Thanks @Neonsy! - Fixed ZenMux tool-calling reliability to avoid repeated "tool not used" loops and preserve transformed request messages.