Qsummary

Qsummary — Session Status & Handoff Summary

Role

Generates a high-density summary of the current session to ensure the user and future agents understand the progress and momentum.

Workflow

Step 1: Data Collection

Gather the following high-signal data:

1. State Stats: Read .qe/state/unified-state.json for token usage and tool call counts.
2. Completed Tasks: Scan .qe/tasks/completed/ for recently finished TASK_REQUEST titles.
3. Semantic Context: Read .qe/context/sessions/{sid}/SNAPSHOT_SUMMARY.md for the active sid (if available) for the current "vibe" and technical state. The 8-char sid comes from the [Session] sid:XXXXXXXX marker in SessionStart additionalContext.

Step 2: High-Density Summary Generation

Use the Haiku model to synthesize the data into a maximum of 5-7 lines. Focus on:

• What: Core achievements and modified files.
• Why: The architectural reason behind the changes (The "Philosophy").
• Next: Immediate next steps for the following session.

Step 3: Output Format

Display the summary in a clean markdown block:

## 🍎 Session Summary [Date]

- **Achievements**: [What was done]
- **Architectural Shift**: [Why/How it changed]
- **Current State**: [Token status / Momentum]
- **Next Steps**: [Pending tasks]

Mandatory Constraints

• Keep it short: No conversational filler or repetitive explanations.
• Accuracy First: Ensure the summary matches the actual state of the .qe/ directory.
• Low Cost: Always use the Haiku tier for this summary.

Will

• Analyze session history and state
• Provide a concise status report
• Focus on technical intent

Will Not

• Re-read every conversation turn (use the state files)
• Write long prose