A native Python CLI agent, Invincat provides AI-powered coding assistance directly within a user's project directory. It targets engineers and power users seeking a terminal-native workflow, offering features like plan-first execution, persistent memory across sessions, and robust guardrails for AI actions, thereby enhancing productivity and reducing context switching.

How It Works

Invincat employs a multi-agent architecture with distinct roles: a Main Agent executes tasks, a Planner Agent facilitates plan review, and a Memory Agent asynchronously curates durable user and project memories. Its core innovation lies in a terminal-native workflow that keeps users within their project directory, coupled with robust guardrails for execution actions (file, shell, network) that are approval-gated by default. Long-context durability is achieved through a novel micro-compression technique and offloading, ensuring operational history remains accessible without blocking interactive latency.

Quick Start & Requirements

• Requirements: Python 3.11+
• Installation: pip install invincat-cli or from source (git clone, pip install -e .).
• Run: Navigate to your project directory (cd ~/my-project) and execute invincat-cli.
• Configuration: After the first launch, run the /model command to configure your LLM provider and API Key.

Highlighted Details

• Terminal-Native Workflow: Enables AI-assisted coding directly within the project directory, eliminating IDE or browser context switching.
• Plan-First Execution: The /plan command allows teams to review and approve AI-generated task checklists before execution, mitigating risks associated with automated actions.
• Long-Context Durability: Features micro-compression (<1ms, LLM-agnostic) and context offloading to manage long conversation histories effectively, preserving operational context.
• Governed Memory System: Maintains persistent user and project memories via memory_user.json and memory_project.json, with a focus on evidence-based persistence, lifecycle management, and inspectability via /memory.
• Extensible Architecture: Supports custom skills, subagents, and MCP tools for adapting the assistant to specific team workflows.
• WeCom Integration: Bridges Enterprise WeChat bot messages into the CLI session for seamless collaboration.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or sponsorship were found in the provided documentation.

Licensing & Compatibility

The project's license is not specified in the README, posing a significant barrier to assessing commercial use compatibility or understanding usage restrictions.

Limitations & Caveats

The project's license is not specified in the README, posing a significant barrier to assessing commercial use compatibility. Invincat also requires configuration of external LLM provider API keys (e.g., OpenAI, Anthropic, Google) to function.