Session-First Architecture & Intelligent MCP Automation
Transform your development workflow with autonomous agentic AI.
Built-in MCP tools, layered processing, vision support, and
intelligent cost optimization for professional developers.
Instead of learning complex commands, just have a conversation:
Intelligent agents that understand context, execute complex tasks autonomously, and route work to specialized AI layers for optimal results.
Everything happens in persistent AI conversations. Save, resume, and organize development sessions with intelligent context management and continuation.
Analyze images, screenshots, diagrams, and UI layouts with AI across all 6 providers. Perfect for debugging visual issues and design analysis.
File operations, code analysis, shell commands, and more through MCP protocol integration. No external dependencies or complex setup required.
OpenRouter, OpenAI, Anthropic, Google, Amazon, Cloudflare. Route different tasks to optimal models for cost efficiency and quality.
Real-time usage tracking, 2-marker cache system, smart model routing, and detailed reporting to optimize costs without sacrificing quality.
Powered by MCP (Model-Centric Programming) protocol for seamless AI-tool integration:
Autonomous AI agents for specialized tasks: code review, debugging, context gathering, and more
Smart file operations with batch editing, line-precise modifications, and undo support
Terminal commands with foreground/background execution and process management
AST-based code search and refactoring with semantic understanding across languages
Intelligent codebase search (works best with octocode MCP for enhanced results)
Research and documentation lookup via Brave Search API integration
Octomind integrates seamlessly with octocode MCP server for semantic codebase search and enhanced code understanding. Get the most out of your AI development workflow.
Rich CLI interface with intelligent commands for autonomous development:
/done
Finalize task with memorization, summarization, and auto-commit
/run <command>
Execute configured layer-based workflows
/layers
Toggle layered processing on/off
/image <path>
Attach images for AI analysis (PNG, JPEG, GIF, WebP, BMP)
/context [filter]
Display session context with filtering (all, assistant, user, tool)
/cache
Mark cache checkpoint for cost savings
/report
Generate detailed usage report with cost breakdown
/info
Display token usage and costs in real-time
/model [model]
View or change current AI model dynamically
Sessions automatically manage context, optimize costs, and provide intelligent continuation. Resume any session exactly where you left off with full context preservation.
All defaults in config-templates/default.toml
. No hardcoded values, complete customization possible.
Developer (full tools) vs Assistant (chat only). Create custom roles for specific workflows.
Route tasks through specialized AI layers: task_refiner → task_researcher → developer. Use cheaper models for simple tasks.
agent_<name>(task="...")
MCP tools route tasks to specialized AI agents for focused, autonomous processing.
/run <command>
executes configured layer-based workflows. /done
finalizes tasks with auto-commit.
Spending thresholds, smart model routing, 2-marker caching, and detailed reporting to optimize costs while maintaining quality.
Provider | Format | Features |
---|---|---|
OpenRouter | openrouter:provider/model |
Multi-provider access, caching, vision models |
OpenAI | openai:model-name |
Direct API, cost calculation, GPT-4o vision |
Anthropic | anthropic:model-name |
Claude models, caching, Claude 3+ vision |
google:model-name |
Vertex AI, Gemini 1.5+ vision support | |
Amazon | amazon:model-name |
Bedrock models, AWS integration, Claude vision |
Cloudflare | cloudflare:model-name |
Edge AI, fast inference, Llama 3.2 vision |
AI agents that understand context and execute complex development tasks without constant supervision.
No indexing, no complex configuration. Install and start autonomous development immediately.
Works best with octocode MCP for semantic codebase search and enhanced understanding.
Analyze screenshots, diagrams, and UI layouts across all 6 AI providers for comprehensive visual debugging.
Route simple tasks to cheaper models, complex analysis to premium models. Maximize efficiency and cost savings.
Create new capabilities, agents, and workflows through configuration alone. No code changes needed.
curl -fsSL https://raw.githubusercontent.com/muvon/octomind/master/install.sh | bash
export OPENROUTER_API_KEY="your_key"
octomind session
# Build from source
cargo install --git https://github.com/muvon/octomind.git
# Download from releases
# See: https://github.com/muvon/octomind/releases
Free and open source software. Use, modify, and distribute freely.
Performance, safety, and reliability. Cross-platform support for macOS, Linux, and Windows.
Contributions welcome! Help improve AI-powered development tools for everyone.