Catnip is an agent harness and orchestrator project under the Weights & Biases (wandb) organization designed to manage, execute, and monitor automated coding agent workflows. In standard agentic pipelines, Catnip acts as the orchestration layer that drives agent-environment interactions, executing task loops and logging execution metadata. Because the primary repository is currently unavailable or restricted, public verification of supported model providers, CLI syntax, and configuration surfaces is limited. The project is positioned around combining developer tool orchestration with the observability and experiment tracking patterns typical of the Weights & Biases software ecosystem. Licensing, model provider compatibility, and ongoing release status depend on repository access under the Weights & Biases GitHub organization.
Agent workflow orchestration — Manages automated agent cycles and code modification tasks. - Experiment tracking integration — Aligns agent task execution with Weights & Biases observability tooling. - Harness infrastructure — Provides scaffolding to execute, evaluate, and benchmark autonomous developer agents. - CLI-driven execution — Designed for terminal and headless CI/CD evaluation pipelines.
Use cases
Use Case: Agent Evaluation and Benchmarking Developers testing coding agents can use Catnip to orchestrate reproducible benchmarks across codebases and track step-by-step performance metrics.
Use Case: Automated Task Execution Engineers can run multi-step code refactoring or bug-fixing routines where an orchestration harness coordinates tool execution and file modifications.
Who it is for
Catnip is aimed at AI engineers, ML researchers, and developer tooling teams building or benchmarking agentic coding systems. It is not intended for end users seeking a production-ready, out-of-the-box interactive terminal assistant.
Install guide
Installation Note: The upstream repository wandb/catnip is currently inaccessible or private. Standard Python installation from source follows this pattern:bash git clone https://github.com/wandb/catnip.git cd catnip pip install -e .
First Run Configure required environment variables and API keys: bash export WANDB_API_KEY="your-wandb-api-key" export OPENAI_API_KEY="your-model-api-key"
Example Session Run an agent task harness via the CLI interface: bash catnip run --task "Generate unit tests for src/api.py"
Unship — Unship is an open-source agent harness and plugin integration framework developed by mbenhard. It is designed to package and orchestrate…
Skillreaper — Skillreaper is an open-source skill lifecycle management and auditing tool for AI agent ecosystems, developed by thousandflowers. It analyzes agent…
Stratless — Stratless is an open-source agent harness and execution orchestrator developed by the Stratless AI team. It provides a modular infrastructure…
Playwright Praman — Playwright Praman is an open-source test harness and agent orchestrator developed by Maheshwar Kanitkar. It is designed to bridge terminal…
Zosma QA — Zosma QA is an open-source testing-focused agent harness and orchestration framework created by Zosma AI. It provides scaffolding, structured prompt…
Agent Island — Agent Island is an open-source agent orchestration harness created by developer tristan666666 on GitHub. Designed as a framework for running…