Valo Toolkit — Free Resources for AI Development
The standards, workflows, and specialist agents we use to build Valo. Free to copy, adapt, and use in your own projects.
Start with the guide, copy the five agents into your project, or use both. Everything here works without installing Valo.
How to use this
> building something new?
- give your assistant the style guide
> reviewing what was built?
- run the relevant agents — or all five if you are lazy
> existing project is already messy?
- improve touched areas, or migrate one boundary at a time
Ask your assistant to adapt the guide and agents to your project’s stack, commands, risks, and exceptions. Keep the universal rules; add a small project-specific overlay.
AI-Optimized Code Quality
The universal style and architecture standard: ownership, boundaries, state, runtime contracts, maintainability, and proof.
Quality Pipeline
Five focused agents that enforce the standard: review, durability, organization, tidying, and validation.
Code Reviewer Agent
Reviews current correctness, security, supported contracts, and runtime boundaries.
Durability Agent
Finds future breakage in shared state, lifecycle, persistence, timing, and external drift.
Organizer Agent
Audits module ownership and dependency shape, then applies only bounded structural fixes.
Tidier Agent
Improves local clarity and consistency without changing behavior or public contracts.
Validator Agent
Runs risk-matched evidence and reports passed, failed, unavailable, and unproved checks honestly.
Prompt Engineer Agent
Reviews CLAUDE.md files, agent prompts, and documentation for clarity, AI-optimization, and Claude 4.x best practices.
Web Researcher Agent
Advanced search craft, anti-SEO filtering, multilingual search, and query decomposition for real answers.
Concept Pages (Valo Memory)
Document what broke, not what works. Scar-tissue knowledge that triggers automatically when AI touches related files.