Coding

What This Skill Does

The Coding skill provides OpenClaw with persistent, user-specific memory regarding programming conventions and architectural preferences. Unlike general-purpose AI models that guess or infer preferences based on the current codebase, the Coding skill relies exclusively on a consent-first architecture. It functions as a local repository of your "tribal knowledge," storing preferences in ~/coding/memory.md to ensure every code generation aligns with your specific stack requirements and stylistic standards. By maintaining a strict 100-line limit and enforcing a 5-word-per-entry rule, it prevents memory bloat while keeping your development workflow consistent across sessions.

Installation

To integrate this skill into your environment, use the OpenClaw command-line interface. Run the following command in your terminal:

clawhub install openclaw/skills/skills/ivangdavila/coding

Once installed, ensure your environment supports local file writes by initializing the memory directory: mkdir -p ~/coding. The system will automatically create memory.md and history.md upon your first confirmed preference.

Use Cases

• Standardization: Ensure all generated Python code uses type hints or specific naming conventions (e.g., camelCase vs snake_case).
• Stack Enforcement: Force the agent to default to specific libraries, such as using Pydantic for validation or opting for Tailwind CSS over standard CSS modules.
• Project Structure: Maintain consistent folder patterns, such as keeping tests colocated with source code rather than in a separate top-level directory.
• Constraint Management: Explicitly ban certain antipatterns or libraries that you prefer to avoid in your project environment.

Example Prompts

1. "I prefer to use functional components in React and avoid class-based structures. Should we remember this?"
2. "Show me all my current coding preferences stored in memory."
3. "Forget the previous preference about using tabs for indentation, I want to switch to 2-space indentation."

Tips & Limitations

• Precision is Key: Keep entries under 5 words for maximum clarity. Use specific triggers like 'naming:', 'structure:', or 'stack:' to organize your thoughts.
• Avoid Overloading: The 100-line limit is a feature, not a bug. Archive outdated preferences to history.md to maintain high-quality context.
• Privacy: This skill never observes your project files directly. You must explicitly teach it. If the agent makes an assumption you dislike, simply correct it to initiate the 'Should I remember?' prompt, allowing you to curate your coding identity actively.