Skip to content

Sessions

What is a session?

A RaiSE session maps 1:1 to your AI assistant session — one Claude Code conversation, one Kimi session, one agent run. When you open a session, RaiSE loads everything your AI partner needs to work effectively in your project:

  • Governance — your principles, requirements, and guardrails
  • Memory — patterns learned from previous sessions, calibration data
  • Context — current epic, story, pending work, deadlines
  • Health checks — environment, MCP servers, graph freshness, pending updates

Without a session, your AI starts from zero every time. With a session, it picks up where the last one left off.

Why sessions matter

Sessions solve the biggest problem of AI-assisted development: amnesia. Every time you start a new conversation with an AI, it forgets everything it learned. Sessions fix this by:

  1. Loading context — your AI knows what you're working on, what patterns apply, and what rules to follow
  2. Tracking work — everything that happens gets recorded in a session journal (decisions, blockers, discoveries)
  3. Persisting learnings — patterns discovered during work survive the session and feed future ones
  4. Building velocity — Session 1 is slow because everything is new. Session 50 is fast because patterns, calibration, and governance have accumulated

This is the compounding effect at the heart of RaiSE: every session makes the next one better.

How to use sessions

Open

In your AI assistant (Claude Code), invoke:

/rai-session-start

This loads your project context, runs health checks, and proposes what to work on based on pending items and recent history.

Work

Use skills to structure your work:

  • Story work: /rai-story-start → /rai-story-design → /rai-story-plan → /rai-story-implement → /rai-story-review → /rai-story-close
  • Bug fix: Start the bugfix pipeline for your issue
  • Exploration: Ask Rai questions, query the graph (rai graph query "topic"), review code

Everything you do is tracked in the session journal. Patterns you discover are persisted to memory.

Close

When you're done working:

/rai-session-close

This reflects on what happened, captures patterns, and records session data for continuity.

What's next