version: "1.0.1" name: rai-session-close description: Capture session outcomes and update memory. Use to close a working session. model: haiku

allowed-tools:

• Read
• Write
• Grep
• Glob
• "Bash(rai:*)"

license: MIT

metadata: raise.work_cycle: session raise.frequency: per-session raise.fase: "end" raise.prerequisites: "" raise.next: "" raise.gate: "" raise.adaptable: "true" raise.version: "5.0.0" raise.visibility: public raise.inputs: |

• session_id: string, required, previous_skill

raise.outputs: |

• session_record: file_path, file
• patterns: list, cli

Session Close

Purpose

Close a working session in ONE composite call. All writes (session record, patterns, corrections, journal, state) happen atomically in raise-cli — this skill only composes the structured summary and presents the result (ADR-093 / ADR-024).

Mastery Levels (ShuHaRi)

• Shu: Walk through each CloseInput field with the developer
• Ha: Compose autonomously, confirm summary before closing
• Ri: Compose and close; surface only the confirmation

Context

When to use: At the end of every working session with meaningful work.

When to skip: No meaningful work happened (plain rai session close clears state).

Inputs: Session activity (this conversation), active session id.

Steps

Step 1: Compose the close state (judgment — this is the LLM's job)

Reflect on the session and build a CloseInput JSON object:

{
  "summary": "1-2 lines: what happened this session",
  "session_type": "feature|research|kata|ideation|maintenance",
  "outcomes": ["merged S7884.1", "created RAISE-7959"],
  "patterns": [
    {"content": "Action + context + reason, 100-300 chars",
     "context": "comma,separated,keywords", "type": "process"}
  ],
  "corrections": [
    {"what": "what the developer corrected", "lesson": "what to do next time"}
  ],
  "next_session_prompt": "Guidance for your future self — highest-signal continuity",
  "narrative": "Optional: decisions, blockers, open threads"
}

Quality bar: patterns are insights worth recalling across sessions (not narrative); next_session_prompt is the single most valuable field.

Step 2: Close (one call)

Use the raise_session_close_full MCP tool with state_json="{the JSON above}", cwd="{project_or_worktree_path}". If MCP tools are not available, fall back to:

# Write the same object as YAML/JSON to a file, then:
rai session close --state-file /tmp/session-output.yaml --project .

Telemetry is emitted server-side — do NOT call signal/topic tools here.

Step 3: Confirm

Present the close card verbatim from the tool response — do NOT recompose these fields:

Session closed — {response.session_id}
Recorded: {response.patterns} patterns, {response.corrections} corrections
Next session: {response.next_session_prompt}

If session artifacts (memory/state files) are tracked by the repo and dirty, offer a chore(session) commit — never mix with story commits.

Output

Quality Checklist

• [ ] ONE composite call — no separate telemetry/memory/journal commands
• [ ] Patterns follow action+context+reason, 100-300 chars
• [ ] next_session_prompt filled — continuity is the point of closing well

References

• Service: raise_cli/session/close.py::process_session_close · Complement: /rai-session-start