by Emilio Osorio

RaiSE v2.2: 37 Skills, MCP Servers, and a New Home for the Website

v2.2 brings 37 skills, MCP server management, multi-agent support, quality and architecture reviews, and the website now lives alongside the source code.

releasev2.2announcement

RaiSE v2.2

We’ve been busy. Here’s what’s new.

37 Skills (up from 20)

The skill ecosystem nearly doubled. New additions include:

  • Quality Review (/rai-quality-review) — catches what linters miss: semantic bugs, type lies, test waste
  • Architecture Review (/rai-architecture-review) — evaluates proportionality using Beck’s four rules
  • Research (/rai-research) — epistemologically rigorous investigation with evidence catalogs
  • Debug (/rai-debug) — root cause analysis using lean methods (5 Whys, Ishikawa)
  • Doctor (/rai-doctor) — self-diagnostics for your RaiSE setup
  • Problem Shape (/rai-problem-shape) — shape vague ideas into well-formed problem statements
  • Bugfix (/rai-bugfix) — formal 6-phase bug fix lifecycle
  • Publish (/rai-publish) — structured release workflow with quality gates

Every skill follows the same contract: purpose, steps, verification, quality checklist. No magic prompts — just structured process.

New CLI Command Groups

Six new command groups expand what the CLI can do:

  • rai backlog — create, search, transition, and comment on issues (Jira integration)
  • rai mcp — manage MCP servers (list, health, tools, install, scaffold)
  • rai gate — discover and check workflow gates
  • rai adapter — list and validate adapter contracts
  • rai release — quality gates and publishing pipeline
  • rai docs — publish and search documentation targets

MCP Server Management

RaiSE now manages MCP (Model Context Protocol) servers as first-class citizens. Install servers from npm/PyPI, health-check them, and call their tools — all from the CLI.

rai mcp install @anthropic/mcp-server-github --type npx --name github
rai mcp health github
rai mcp tools github

Multi-Agent Support

Multiple AI agents can work on the same project simultaneously. Each agent gets isolated session state — no more context leaks between Claude Code, Cursor, and Windsurf instances.

Knowledge Graph Improvements

The graph now covers governance, memory, work tracking, code structure, and identity — with incremental builds and drift detection. Query anything:

rai graph query "testing patterns" --types pattern,guardrail --limit 5

Website Moves Home

This website now lives in the same repository as the framework source code. This means documentation stays accurate — the AI writing the docs can read the code it’s documenting.

Get started: pip install raise-cli && rai init

Full changelog: GitHub Releases