usewhale/Whale
Whale — blazingly fast, terminal-first AI coding agent for DeepSeek. ~98% prompt cache hit rate, 1M context, MCP tools, dynamic workflows.
What it solves
Whale is an AI coding agent designed to provide a fast, cost-effective way to automate coding tasks. It solves the problem of high API costs and slow response times in AI-assisted coding by aggressively reusing cached context, specifically optimized for DeepSeek's long-context models.
How it works
Whale operates as a tool that can be used in the terminal (TUI), as a one-shot CLI command, or in a headless mode for CI/CD. It leverages DeepSeek's 1M token context window and tool-calling capabilities to read files, run commands, and edit code. It also supports the Model Context Protocol (MCP) to connect to thousands of external tools, and allows users to write JavaScript scripts to orchestrate multi-agent workflows (e.g., fan-out research or multi-perspective reviews).
Who it’s for
It is intended for developers who want an AI agent that integrates into their existing environment (terminal, desktop, or CI) rather than replacing their IDE, and those who specifically use DeepSeek models for their AI-assisted coding.
Highlights
- DeepSeek-native: Optimized specifically for DeepSeek's caching, tools, and pricing.
- High Cache Hit Rate: Achieves ~98% prompt cache hit rate to reduce costs.
- Dynamic Workflows: Supports JavaScript-based multi-agent orchestration, compatible with Claude Code.
- Extensibility: Integrates with MCP servers, community skills, and custom plugins.
- Flexible Interfaces: Available as an interactive TUI, a CLI for quick questions, and a headless mode for automated tasks.