AnsibleClaw is a skill generation framework that converts Ansible modules into portable AI agent skill packages, bridging Ansible's 3000+ modules to AI agents like Cursor, Claude Code, and Gemini CLI. The project is classified across multiple domains including Ansible Management, Inventory Management, Playbook Execution, Web Interface, Automation, Orchestration, Configuration Management, Remote Execution, Deployment, and Server Control.
The core functionality of AnsibleClaw operates in two phases. At build-time, the tool runs the ansibleclaw command to resolve Ansible module documentation through ansible-doc, with Galaxy fallback when collections are not installed locally, and emits complete skill packages. If Ansible Automation Platform is configured through environment variables or a .ansibleclaw.yml file, generation can embed Controller defaults including URL, inventory, credential, project, and organization into the skill without baking in bearer tokens. At runtime, generated skills teach AI agents both CLI mode using standard ansible and playbook workflows with ansible-core on the control node, and AAP mode that launches ad-hoc commands and job templates via scripts/aap_run.py using AAP_CONTROLLER_URL and AAP_CONTROLLER_TOKEN environment variables.
The repository includes a comprehensive web dashboard accessible via ansibleclaw ui at http://localhost:8600. The dashboard features a collapsible left sidebar organized into five main sections. The Explore section allows users to browse available Ansible modules by keyword and namespace, view parameters and examples inline, manage installed collections with one-click install and uninstall, and upload starter pack archives to preview playbooks. The Build section enables users to generate single-module skill packages with live SKILL.md preview and multi-target output options, as well as compose multi-module composite skill packages combining several Ansible modules into single use-case-driven skills. The Skills section displays all available skills including built-in and generated ones, allows reading SKILL.md content, downloading as ZIP, and installing to various AI platforms with one click. The Deploy section provides Dev/Test capabilities with playbook selection, inline editing, AI-powered refinement via OpenAI-compatible endpoints, and one-click ansible-playbook execution in dry-run or apply mode, plus Production integration with AAP Controller for project and job template status monitoring. The Agents section includes a Gemini CLI launcher that runs Google's Gemini CLI agent via ttyd with configurable working directory and session management.
Each generated skill package is a complete Agent Skills package containing several key files. The SKILL.md file serves as the primary documentation that agents read to learn module parameters and usage patterns. Scripts include run.sh for wrapping ansible CLI with safe defaults, check.sh for validating prerequisites, publish_playbook.sh for syncing files into AAP Project repositories, and aap_run.py for launching commands via AAP Controller API. The assets directory contains a playbook.yml with example tasks and requirements.yml declaring collection dependencies.
The CLI provides several key commands. The ansibleclaw generate command creates full skill packages for any Ansible module with options to install directly to cursor, claude, or gemini, write to custom directories, or create ZIP archives. The ansibleclaw search command discovers modules by keyword with namespace filtering and detailed JSON documentation options. The ansibleclaw uninstall command removes skill directories from agent platform install paths.
AnsibleClaw ships with four built-in skills always available without repository checkout: ansible_manager teaches AI to run any Ansible module via CLI or AAP, ansible_search enables module discovery via ansible-doc, ansible_skills_factory allows AI to generate new dual-mode skills on-demand, and ansible_aap_guide instructs on using AAP as the production execution platform. Installation requires pip install ansible-claw for core CLI functionality, with optional extras for the web dashboard and development tools. Runtime on target systems requires only ansible-core. AAP integration is enabled through environment variables including AAP_CONTROLLER_URL, AAP_CONTROLLER_TOKEN, and optional configuration for SSL verification, default inventory, credential, and organization settings, supporting both AWX and Red Hat AAP Controller.
