Skip to content

Release 0.26.0

Choose a tag to compare

@github-actions github-actions released this 28 Apr 15:16
· 307 commits to master since this release
eeca7d7

πŸš€ What's Changed

πŸ“‹ Release Summary

This release introduces expanded AI model support for Featherless, NVIDIA NIM, Groq, and BytePlus, alongside enhanced multimodal capabilities including clipboard image pasting and optimized media previews (6cf59b3, 4dba526, 494df4d, 287baeb). The session experience is significantly improved through a refined skill persistence system during context compression, unified MCP tool management, and new real-time performance metrics for token usage and throughput (333248d, 3e4dded, 92efb57, 42abaa4). System stability is further bolstered by streamlined configuration validation, more robust skill lifecycle management, and several fixes to tool parameter handling and session naming (77fd2b3, f6464a3, 5d69774).

✨ New Features & Enhancements

  • provider: add featherless support via octolib 6cf59b30
  • chat: add token usage averages to info command 92efb57a
  • chat: add throughput metrics to info command 42abaa45
  • session: enhance skill persistence during compression 333248d5
  • core: enhance config validation and session naming 77fd2b3f
  • octolib: support Groq, BytePlus, and new AI models e4ed32f3
  • proctitle: add process and terminal title support 3cb24a2f
  • chat: add inline image previews for clipboard paste 494df4d9
  • chat: intercept paste events for media files 0840301b
  • chat: add Ctrl+V support for clipboard media attachments 287baeba
  • add nvidia nim provider support 4dba526f
  • skill: emit skill lifecycle events via WebSocket 7fd5eaf5
  • mcp: unify server management and add tools retrieval 3e4dded7
  • config: support mcp-*.toml override configs and fix auto-bind tracking eb1e2433
  • agents: generalize developer and add skill system 687fd836
  • chat: preserve skills during compression e1f7517e

πŸ”§ Improvements & Optimizations

  • skill: remove forced compression on skill forget ddcbb255
  • core: migrate layers to ACP protocol and update docs fe8c245a
  • session: shorten and reorder session name components 13a1ca9a
  • session: downscale inline image previews 321d00e8
  • acp: replace save and cache with skill command 004c56dc
  • config, layers, session: restructure layers to roles architecture d529770f
  • website: remove website and deployment logic 9467f35f

πŸ› Bug Fixes & Stability

  • config: require command field for ACP layers 0d99d363
  • mcp: ensure MCP function parameters have type field 5d697741
  • mcp: honor dynamic server enable/disable per session ae78b3d3
  • skill: prevent duplicates and fix validation f6464a32
  • session: prevent mid-loop tool compression ff66fd8b

πŸ“š Documentation & Examples

  • session: clarify skill lifecycle and compression 557e42e0
  • pipeline: expand provider, compression, and workflow documentation 6a1e2a4d
  • map: sync project map and architecture docs 23996626

πŸ”„ Other Changes

3 maintenance, dependency, and tooling updates not listed individually.

πŸ“¦ Installation

Quick Install Script (Universal)

curl -fsSL https://raw.githubusercontent.com/Muvon/octomind/main/install.sh | sh

Works on: Linux, macOS, Windows (Git Bash/WSL/MSYS2), and any Unix-like system

Manual Download

Platform Architecture Download
Linux x86_64 (static) octomind-0.26.0-x86_64-unknown-linux-musl.tar.gz
Linux ARM64 (static) octomind-0.26.0-aarch64-unknown-linux-musl.tar.gz
Windows x86_64 octomind-0.26.0-x86_64-pc-windows-msvc.zip
Windows ARM64 octomind-0.26.0-aarch64-pc-windows-msvc.zip
macOS x86_64 octomind-0.26.0-x86_64-apple-darwin.tar.gz
macOS ARM64 octomind-0.26.0-aarch64-apple-darwin.tar.gz

Using Cargo (from crates.io)

cargo install octomind

Using Cargo (from Git)

cargo install --git https://github.com/Muvon/octomind

Verify Installation

octomind --version