AI coding agents learn your project conventions, parameters, and lessons during a session. Then the session ends and it's all gone. Switch to a different agent and you re-explain everything.
stato captures what the agent learned into validated Python modules on disk:
stato crystallize: agent writes structured .stato/ modules (skills, plan, memory, context)
stato validate: 7-pass compiler checks before writing to disk
stato bridge: generates CLAUDE.md, .cursorrules, or AGENTS.md that agents read automatically
Four transfer paths: across sessions (files persist on disk), across projects (snapshot/import archives), across platforms (Claude Code/Cursor/Codex from one source), and from web AI to coding agents.
Also: composition algebra (snapshot, slice, graft, merge), privacy scanner (19 patterns catch API keys before sharing), package registry, and resume after /compact.
A/B tested: asked the same question with and without .stato/ files. With stato, agents gave detailed project-specific answers. Without, they didn't recognize the tools.
prism2•1h ago
stato captures what the agent learned into validated Python modules on disk: stato crystallize: agent writes structured .stato/ modules (skills, plan, memory, context) stato validate: 7-pass compiler checks before writing to disk stato bridge: generates CLAUDE.md, .cursorrules, or AGENTS.md that agents read automatically
Four transfer paths: across sessions (files persist on disk), across projects (snapshot/import archives), across platforms (Claude Code/Cursor/Codex from one source), and from web AI to coding agents.
Also: composition algebra (snapshot, slice, graft, merge), privacy scanner (19 patterns catch API keys before sharing), package registry, and resume after /compact.
A/B tested: asked the same question with and without .stato/ files. With stato, agents gave detailed project-specific answers. Without, they didn't recognize the tools.