Generate Claude Code Context Artifacts with aidd-context

Where this fits

You've been using a slash command for a while, but now you want to promote it into a proper skill with a router, atomic actions, and evals. Hand-writing all that boilerplate is tedious and error-prone. That's exactly when aidd-context:03:context-generate steps in—it captures your intent and generates the entire skill structure, or refactors an existing one, without you touching every file manually.

Why agents benefit

• Consistent structure: The skill enforces the router-based pattern (SKILL.md + atomic actions + minimal evals) so every generated skill follows the same tested architecture.
• Intent capture before implementation: It prompts you to clarify output and decide generate-vs-modify, preventing scope creep and reducing rework.
• Eval-first development: Evals are declared upfront, making it easy to verify each action works in isolation—something you'd likely skip if writing by hand.
• Safe refactoring: Adding or removing actions becomes a guided process, with the generator updating the router and evals automatically.
• Slash command migration: Converts an ad-hoc slash command into a maintainable, testable skill without losing the original behavior.

Practical scenarios

Migrating a slash command to a skill

You have a /review slash command that runs a series of prompts. Instead of leaving it as a monolithic script, you invoke the generator, describe the intent, and get a SKILL.md router plus individual actions for each review step, each with its own eval.

Adding a new action to an existing skill

Your skill has grown, and you need to add a "summarize changes" action. The generator asks whether you're modifying or generating, then scaffolds the new action file, wires it into the router, and suggests a minimal eval.

Generating a new agent or rule

You want to add a specialized agent for code review or a rule for commit message formatting. The generator provides a template-based approach, ensuring the new artifact matches your project's conventions and integrates cleanly.

Add it to your agent workflow

Install via npx:

npx skills add ai-driven-dev/framework --skill aidd-context:03:context-generate

Then, in Claude Code, invoke the skill with a prompt like:

Use aidd-context:03:context-generate to create a skill that validates JSON files in the project.

The generator will ask clarifying questions, then produce the complete skill directory with router, actions, and evals ready for review. For more context on where this fits in the ecosystem, see this skill's page.

Compare before adopting

Before committing to this skill, evaluate it against alternatives. Look for:

• Quality signals: Are the generated artifacts idiomatic and well-tested? Check the framework's own evals and examples.
• Maintenance freshness: Is the repo actively updated? The 455 stars suggest community interest, but verify recent commits.
• Alternatives: Could you use a simpler templating script or a different framework? Consider whether the router-based pattern is necessary for your project.
• Workflow fit: Does the generator's decision tree match how you work? It's designed for generating whole skills, not editing single actions—know its boundaries.

Why it is worth tracking

This skill is part of a larger framework that emphasizes eval-first, testable agent components—a pattern gaining traction in the AI engineering community. With 455 stars and a clear focus on maintainability, it's a strong candidate for teams standardizing their Claude Code workflows. Even if you don't adopt it today, watching its evolution can inform how you structure your own agent artifacts. Evaluate it when you're ready to move from ad-hoc commands to disciplined, testable skills.