nanoclaw is an open-source CLI agent framework and containerized execution harness developed by Nanoco AI. Designed around Claude-powered workflows and headless container environments, it orchestrates automated coding, testing, and file operations within isolated sandboxes. In standard workflows, nanoclaw executes tasks through a containerized pipeline that coordinates agent instructions, project documentation, and custom skills. It integrates deeply with Anthropic's Claude models, parsing configuration contracts (such as CLAUDE.md, AGENTS.md, and structured skill definitions) to direct code generation, automated refactoring, and headless CI runs. What distinguishes nanoclaw is its focus on safe, containerized execution coupled with modular agent persona and skill configuration. By isolating agent operations inside container runtimes while maintaining synchronized host-side repositories and memory settings, it mitigates unapproved system modifications and supports reproducible autonomous workflows. nanoclaw is distributed as an open-source TypeScript project under active development, free to use with the operator providing their own Anthropic API access and container runtime environment.
Containerized agent runtime — Executes agent scripts and tasks inside isolated container environments to prevent uncontrolled host modifications. - Claude-native configuration — Parses CLAUDE.md and AGENTS.md project specifications alongside structured skill definitions to direct agent actions. - Skill and persona orchestration — Supports modular skill mappings and persona compositions configured for specific repository tasks. - Atomic file migrations — Uses verified atomic file writes and staging pipelines to ensure clean repository state transitions. - TypeScript host harness — Built in TypeScript with integrated test suites for both host orchestration and container execution. - Automated update pipelines — Features dedicated transaction-gated update mechanics to handle breaking configuration migrations safely.
Use cases
Use Case: Sandboxed Autonomous Refactoring A developer needs to perform repository-wide refactoring or dependency upgrades across a complex TypeScript codebase. Using nanoclaw, the agent runs the transformations, builds, and test suites inside a containerized sandbox before staging verified changes back to the host repository.
Use Case: Headless CI Code Maintenance Teams running automated issue-to-PR pipelines deploy nanoclaw in headless CI environments. The agent consumes repository-specific rules from CLAUDE.md, applies targeted bug fixes or symlink migrations, and passes local lint and unit test suites autonomously.
Use Case: Skill-Augmented Agent Tasks A developer defines custom skills and personas for domain-specific tasks such as API migrations or schema updates. nanoclaw loads the selected skills into the container environment and executes multi-step terminal workflows using Anthropic models.
Who it is for
nanoclaw is intended for software engineers and DevOps teams who require container-isolated, autonomous coding agents configured with Claude models and structured skill files. It is less suited for developers seeking an interactive in-IDE chat sidebar or those looking for a zero-configuration GUI tool without Docker or container dependencies.
Install guide
Installation Clone the repository and install dependencies using Node.js and npm: bash git clone https://github.com/nanocoai/nanoclaw.git cd nanoclaw npm install npm run build
First Run Ensure Docker or your local container runtime is running, then configure your Anthropic API credentials in your environment: bash export ANTHROPIC_API_KEY="your-api-key-here" Verify that project configuration files (such as CLAUDE.md or .agents/skills) are present in your target repository root.
Example Session Run the nanoclaw runner against your workspace to execute automated agent checks or tasks: bash npm test npx tsx src/index.ts --task "Refactor compose pipelines to use atomic writes"
Nullclaw — Nullclaw is an open-source, lightweight AI coding assistant and agent harness designed to operate with minimal overhead. Developed using Zig…
Hermes Agent — Hermes Agent is an open-source terminal-based AI coding agent developed by NousResearch. Designed to bridge terminal workflows and modern IDE…
Orkas — Orkas is an open-source desktop agent harness and workflow orchestrator developed by Orkas-AI. Built to bridge autonomous AI execution with…
SWE-agent — SWE-agent is an open-source autonomous software engineering agent harness built and maintained by researchers from Princeton University and Stanford University.…
Stratless — Stratless is an open-source agent harness and execution orchestrator developed by the Stratless AI team. It provides a modular infrastructure…
Agentic Engineering Framework — Agentic Engineering Framework is an open-source orchestration harness and governance layer developed by Dimitri Geelen. It is designed to constrain,…