v1.1.0
AI-RPI Protocol v1.1.0
This release focuses on adaptive governance, specialist workflows, and cleaner repo structure for day-to-day agent usage.
Highlights
- Added a Codex IDE adapter (
adapters/ides/codex.md). - Added major core system docs: progressive loading, deterministic guardrails, validation model, artifact ladder, evaluation flywheel, operational units, role-based collaboration, setup lifecycle, and more.
- Added an ultra-light protocol profile for low-risk requests.
- Expanded specialist agents and skills for research, challenge, implementation, verification, and multi-lens code review.
- Added PRD/Tech Spec templates and project-info generation templates under
templates/project-info-generation/. - Added public launch assets and benchmark artifacts.
Behavior Changes
- Startup behavior is now mandatory across entry surfaces (
AGENTS.md,CLAUDE.md, protocol entry files). - Runtime behavior is formalized around Mode (
Explore,Discuss,Review,Patch,Feature,Build), Depth (minimal,balanced,full), and Persona. - Code-review routing was hardened: explicit review/audit requests now route to
skills/code-review.md. - Project context handling now prioritizes durable
memory.mdand project-scopeduser-preferences.md.
Repository Cleanup
- Removed legacy experimental workspace files from the codebase (backend plan doc retained).
- Migrated old
core/project-info/*surfaces totemplates/project-info-generation/*.
Credits
- Thanks to @fabioasdias for PR #3 ("Fix typos and enhance clarity in README").
Full changelog: https://github.com/MiguelAxcar/ai-rpi-protocol/blob/v1.1.0/CHANGELOG.md