Skills Laravel Package

Product Decisions This Supports

• AI Agent Integration Roadmap: Enables modular, dependency-managed AI skill distribution for coding agents (e.g., Claude, Cursor, Aider). Accelerates adoption of agent-assisted workflows by reducing friction in skill onboarding.
• Build vs. Buy: Avoids reinventing skill distribution infrastructure; leverages Composer’s existing ecosystem for dependency management.
• Use Cases:
  • Team Onboarding: Standardize AI skills across engineering teams via Composer packages.
  • Agent Agnosticism: Share skills across multiple agents (e.g., .agents/skills → .claude/skills).
  • Security: Explicit trust controls mitigate prompt-injection risks from untrusted packages.
  • DevOps Integration: Auto-sync skills on composer install aligns with CI/CD pipelines.

When to Consider This Package

• Adopt if:

  • Your team uses multiple coding agents (e.g., Claude, Cursor, Aider) and needs shared skill repositories.
  • You want Composer-managed AI skills (versioned, dependency-tracked) instead of manual file copies.
  • Security is critical: Explicit trust controls are required for untrusted environments.
  • You’re building a platform for AI-assisted development and need agent-agnostic skill distribution.

• Look elsewhere if:

  • Your use case is single-agent (e.g., only Claude) and doesn’t need cross-agent compatibility.
  • You prefer centralized skill hosting (e.g., GitHub repos, private registries) over Composer.
  • Your team lacks PHP/Laravel infrastructure (package requires Composer).
  • You need dynamic skill loading (e.g., runtime updates without composer install).

How to Pitch It (Stakeholders)

For Executives: "This package lets us distribute AI skills (e.g., coding guidelines, templates) as Composer dependencies—just like any other library. Teams can auto-sync skills on composer install, reducing onboarding time and ensuring consistency across agents like Claude or Cursor. It’s secure (trust controls), scalable (versioned), and integrates with our existing PHP/Laravel stack. Think of it as ‘npm for AI skills’—but for developers."

For Engineering: *"llm/skills solves the ‘how do we share AI skills across agents?’ problem by treating them as Composer packages. Key benefits:

• One source of truth: Skills live in .agents/skills/ and are mirrored to agent-specific paths (e.g., .claude/skills) via symlinks/junctions.
• Security: Only trusted packages (explicit or via composer.json deps) are synced; malicious payloads are blocked.
• DevOps-friendly: Auto-syncs on composer install (configurable) and works in CI/CD.
• Agent-agnostic: Works with Claude, Cursor, Aider, etc., without duplication. Example: Add composer require llm/skills to your project, and skills from packages like acme/skills-refactor auto-populate into .agents/skills/refactor/."*

For Security/Compliance: *"The package enforces strict trust controls: skills are only copied from:

1. Direct Composer dependencies (require/require-dev),
2. Explicitly trusted packages (configured in skills.json), or
3. CLI-whitelisted packages. Malformed or untrusted packages are skipped with warnings. Symlinks/junctions ensure no duplicate files, and the target directory is configurable for isolation."*