Keen Code

Keen Code is a coding assistant extension that classifies tasks by complexity and routes them to appropriate chat models.

Keen Code classifies accepted REPL user turns by task complexity using a configured decision model. It evaluates conversation state asynchronously alongside normal response generation while excluding commands and failed submissions. The decision model answers three questions regarding category, consequentiality, and new-task probability. An optional model router consumes these signals to decide whether to switch among configured simple, standard, and complex chat models based on specific routing policies. If evaluators are unavailable or timeouts occur, the system fails open without blocking the user turn.

Keen Code at a glance

Best for: Software engineer
Runs on: CLI
Pricing: Free
Released: October 2026
Category: AI coding tool
Website: mochow13.github.io

Key features

Task classification using decision models
Asynchronous evaluation alongside responses
Optional model router for chat models
State serialization up to 80 KiB
Fail-open behavior during errors
Usage ledger tracking

Use it for

Classifying REPL user turns by task complexity
Routing chat models based on task category
Evaluating conversation state and metadata

Alternatives to Keen Code

OpenCode — Works with 75+ model providers, local models, and existing subscriptions
jevgrep — Reduces coding agent costs by providing targeted, relevant repository context.
OpenClaude — Connects one terminal interface to over 200 different AI models.
Ponytail — Cuts unneeded code and dependencies by enforcing a minimal-code ladder.
ZCode — Integrates GLM-5.3 with multi-agent workflows and remote chat control.
Command Code — Repairs open model tool calls and learns developer patterns automatically.
CodeGraph — It eliminates file crawling, reducing AI agent tool calls by 88% and cutting costs.
SuperClaude — Adds 30 slash commands and 8 MCP servers to Claude Code.
pstack — Improves AI coding agent outcomes with rigorous workflows.
GoRoute — Consumes 5.1MB of RAM with sub-10ms latency for coding agents.
SpiderMonkey — It watches the live machine and steers the agent toward what matters.
Deep Code CLI — Optimized specifically for DeepSeek models with a one million token context.

All Keen Code alternatives