Kodo is an open-source AI agent harness and execution orchestrator developed by Ilya Kamenshchikov (ikamensh). It is built to execute, test, and benchmark multi-step automated coding workflows across diverse large language model backends. Kodo operates primarily through CLI-driven iteration pipelines and Python-based execution harnesses. It interfaces with frontier language models to execute code modifications, run smoke tests, and evaluate generated solutions against structured acceptance criteria. What distinguishes Kodo from conventional single-turn CLI coding agents is its focus on formal iteration cycles, testability policies, and comparative model benchmarking. The framework treats coding tasks as reproducible test-driven loops rather than ad-hoc chat sessions. Kodo is an open-source project hosted on GitHub under active development with public acceptance suites and iteration specifications.
Iterative Workflow Harness — Orchestrates multi-step code generation and refinement loops with structured iteration policies. - Model Benchmarking — Includes benchmarking tools to measure and compare code output quality across different model families. - Testability Enforcement — Integrates acceptance testing, interactive smoke test mocks, and validation checks directly into agent runs. - Multi-Model Interoperability — Configurable harness architecture capable of evaluating multiple frontier reasoning and coding models. - Scriptable CLI Engine — Provides modular Python packages and automation scripts for headless terminal workflows.
Use cases
Use Case: Model Benchmarking Developers and AI engineers can execute standardized coding tasks across different LLM backends to compare accuracy, speed, and iteration efficiency.
Use Case: Test-Driven Iteration Cycles Engineers can run automated loops where Kodo attempts implementation tasks, evaluates test outputs, and iteratively refines the codebase until all acceptance criteria pass.
Use Case: Automated Agent Evaluation Pipelines Tool builders can integrate Kodo scripts into CI/CD pipelines to validate agent behavior and harness policies using hardened smoke test mocks.
Who it is for
Kodo is designed for AI tooling engineers, researchers, and backend developers building and evaluating autonomous coding agent workflows. It is not intended for non-technical users seeking a simple IDE autocomplete or chat-based assistant.
Install guide
Installation Clone the repository and install the package in an active Python environment: bash git clone https://github.com/ikamensh/kodo.git cd kodo pip install -e .
First Run Set the required environment variables for your chosen LLM provider: bash export OPENAI_API_KEY="your-openai-api-key" export ANTHROPIC_API_KEY="your-anthropic-api-key"
Example Session Inspect available orchestrator options and run the harness against an iteration specification: bash python -m kodo --help
OpenCode — OpenCode is an open-source AI coding agent developed by Anomaly that operates across terminal environments, desktop applications, and IDE extensions.…
Plandex — Plandex is an open-source terminal-based AI coding agent designed to handle complex, multi-file engineering tasks in large codebases. Created by…
SWE-agent — SWE-agent is an open-source autonomous software engineering agent harness built and maintained by researchers from Princeton University and Stanford University.…
OpenHands — OpenHands (formerly OpenDevin) is an open-source autonomous AI software development agent and orchestration platform. It provides an extensible runtime environment…
Vibe Kanban — Vibe Kanban, originally developed by Bloop AI, is an open-source orchestration harness and visual task board designed to manage and…
Weaver — Weaver is an open-source agent harness and orchestration framework maintained by sean35mm on GitHub. Built on top of the Bun…