A comprehensive engineering framework for designing libraries, CLIs, and APIs specifically optimized for autonomous AI agent consumption.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ax-development
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install ax-development using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is AX Development Framework?
The AX Development framework (Agent Experience) is a disciplined approach to software engineering that prioritizes the needs of AI agents as primary users. As software is increasingly consumed by LLMs and autonomous entities rather than just humans, this framework provides the principles necessary to ensure code, APIs, and interfaces are reliable, predictable, and discoverable. By utilizing Openclaw Skills to implement these standards, developers can create robust systems that reduce agent hallucinations and operational friction.
At its core, AX focuses on deterministic outputs, machine-readable error contracts, and co-located documentation. These principles ensure that an agent can interact with a module with zero ambient knowledge, parsing structured data instead of ambiguous prose. Integrating these methodologies through Openclaw Skills allows teams to build future-proof infrastructure that scales alongside the growing capabilities of AI agents.
AX Development Framework Use Cases
- Designing new software libraries or APIs intended for integration into AI agentic workflows.
- Refactoring legacy CLI tools to provide structured JSON/JSONL outputs for machine consumption.
- Building deterministic data processing pipelines where output consistency is critical for agent reliability.
- Creating self-documenting codebases where agents can autonomously discover I/O contracts and invariants.
How AX Development Framework Works
- Implement fast-fail validation at the system boundary to reject invalid agent inputs before executing expensive operations.
- Enforce determinism by isolating non-deterministic elements like timestamps or random IDs and making them injectable for testing.
- Adopt a structured error taxonomy (e.g., INVALID_*, MISSING_*) to provide agents with parseable codes and actionable metadata.
- Design narrow, composable primitives that allow agents to use specific pipeline steps independently.
- Co-locate documentation and executable tests within the source tree so agents can find contracts through file exploration.
- Verify the module against the AX Decision Checklist to ensure it requires no implicit environmental knowledge.
AX Development Framework Setup
To begin implementing AX principles in your project using Openclaw Skills, structure your directory to support co-located documentation and machine-readable contracts. Initialize your project structure with the following pattern:
mkdir -p src/my-module
touch src/my-module/mod.ts # Implementation
touch src/my-module/contract.md # Agent-readable documentation
touch src/my-module/mod_test.ts # Executable specifications
Ensure your CLI tools are configured to default to JSON output for agent compatibility:
# Example of an AX-compliant CLI call
my-tool --format json --explicit-flag value
AX Development Framework Data Schema & Taxonomy
The AX Development framework organizes data and metadata to be highly structured and machine-accessible. The following schema defines the typical organization for Openclaw Skills compliant projects:
| Component | Format | Description |
|---|---|---|
| Error Responses | JSON | Includes a stable code and a data object with context. |
| CLI Outputs | JSONL / JSON | Default machine-readable streams with stable keys. |
| Documentation | Markdown | Co-located contract.md files within one directory level of source. |
| Invariants | Tests | Executable specifications that define boundary behavior. |
| Exit Codes | Integer | 0 for success, 1 for user error, 2 for system/environment error. |
AX Development Framework Advanced Features
- Versioned JSON output formats to ensure breaking changes don't disrupt agentic workflows.
- Standardized error taxonomy for consistent cross-project machine parsing.
- Multi-agent support through the use of composable primitives and narrow interface contracts.
- Support for human-readable opt-in flags while maintaining machine-first defaults.
- Automated AX compliance checking via the Decision Checklist integrated into CI/CD.
SKILL.md
Loading
Related Openclaw Skills
TickTick Official CLI Skill
A command-line interface for managing Dida365 projects and tasks using the official Open API and OAuth authentication.
Dida365 (TickTick) Task Manager
A comprehensive CLI-based tool for managing Dida365 tasks and projects through Openclaw Skills.
Tarot Skill
A professional-grade tarot reading module that enables AI agents to conduct deep-dive spreads, provide energy guidance, and generate visual card interpretations.
Agent Team Builder
An interactive framework for designing, configuring, and deploying multi-agent collaboration teams using standardized OpenClaw architecture.
localhost-bridge
A networking utility that bridges Docker containers to host-side localhost services using socat and systemd for seamless AI agent deployments.
Multi-Agent Sandbox
A robust infrastructure for deploying isolated, sandboxed agents that collaborate across gateways via Discord and private networking.
Featured*
