# Compact Session

> Summarize entire session and save to index as compact Q&A pairs

- Skill: `couzip/compact-session` (Agent Skill)
- Install (CLI): `npx skillmds@latest add couzip/compact-session`
- Raw SKILL.md: https://api.skillmd.com/api/skills/couzip/compact-session/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: couzip (https://skillmd.com/u/couzip)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/couzip/compact-session

---


# Compact Session

## CRITICAL RULE

store_memory must be called EXACTLY ONCE. No retries. No corrections.

## Pre-call Checklist (ALL must be YES before calling store_memory)

- [ ] Identified ALL main topics (3-5)?
- [ ] Written summaries for EVERY topic?
- [ ] Combined ALL summaries into ONE content string?
- [ ] Displayed the complete content string?

## Process

1. List all main topics discussed in this session (3-5 topics)
2. For each topic, write: "Human: [question]\nAssistant: [answer]"
3. Concatenate all pairs with blank lines between them
4. Display the COMPLETE content string
5. Call store_memory ONCE with that complete string

## Format Example

Content string should look like:

    Human: How to setup CC-Forever?
    Assistant: Run /cc-forever:setup to configure.

    Human: What embedding models are available?
    Assistant: English (MiniLM), Multilingual (e5), Japanese (Ruri).

## store_memory Call

- content: The COMPLETE string containing ALL topic summaries
- project: current directory name
- tags: ["compact", "session-summary"]

## After Calling

Report: "Saved N topic summaries" and STOP. Do not call store_memory again.