Description
miao is an open-source AI coding agent designed for terminal use, offering a flexible and efficient coding experience. It allows users to select models from various providers, enabling seamless model switching within a session. This flexibility ensures that users can prioritize reasoning quality, speed, or cost without restructuring their workflow around a single vendor.
Read More
The tool supports collaboration by allowing users to delegate tasks to specialist subagents, which can be continued later. This feature is enhanced by the ability to list and message other sessions within the same project, subject to permissions. miao also provides detailed insights into token usage and costs, with per-turn usage metrics, estimated costs, and prompt-cache telemetry.
Durable sessions and an event-backed history ensure that users can list, reopen, fork, and export sessions. Large tool results are managed efficiently, with full outputs stored temporarily when available. miao's capabilities include autonomous continuation, cost budgets, and tuning options for pruning and compaction.
The tool is pre-1.0 and emphasizes its experimental nature, with features like Code Mode and a kernel sandbox available in specific runtimes. miao is ideal for developers looking to integrate AI into their coding workflow, offering a comprehensive suite of tools to enhance productivity and collaboration.
- CategoryAI Agent Builders
- Pricing ModelOpen Source
- DeploymentSelf-hosted
- PlatformsTerminal
miao's Core Features
Open-source AI coding agent
Model selection and switching
Task delegation to subagents
Token usage and cost insights
Durable sessions and history export
Autonomous continuation and cost budgets
Experimental Code Mode
Kernel sandbox compatibility
How to use miao?
Install: Set up miao on your terminal
Configure: Choose models and set permissions
Use: Delegate tasks and manage sessions
Optimize: Monitor token usage and costs
miao's Use Cases
- Model Comparison
- Task Delegation
- Session Management
- Token Monitoring
- Collaborative Coding
