Receber por email
← Voltar para o catálogo

memem-welcome

Re-show the full memem welcome screen regardless of whether the user has seen it before. Use when the user wants to see the intro, tutorial, or welcome again.

34estrelas
Atualizado há 3 meses

Ver no GitHub ↗ Licença: MIT

Como adicionar

/plugin marketplace add TT-Wang/memem

O comando exato pode variar conforme o repositório. Confira o README no GitHub.

Para o autor da skill

Cole no README do seu repo

Mostra que sua skill está catalogada na Skillteca, gera backlink e tráfego rastreável.

[![Skillteca](https://www.skillteca.com.br/api/badge/memem-welcome/svg)](https://www.skillteca.com.br/skills/memem-welcome)

Skills relacionadas

Ver todas de Dados e Análise →

Alerta por categoria

Receba novas skills de Dados e Análise toda segunda

1 email curto, só com as skills novas de Dados e Análise. 4 minutos de leitura, sem spam, cancela com 1 clique.

Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.

Ler descrição completa

Force-display the full memem welcome (the brand-new-user view), even if the user has already dismissed it. This is the "re-open walkthrough" handle — it exists so returning users who want the intro again don't have to delete state files manually.

Step 1 — Print the banner:

  ███╗   ███╗███████╗███╗   ███╗███████╗███╗   ███╗
  ████╗ ████║██╔════╝████╗ ████║██╔════╝████╗ ████║
  ██╔████╔██║█████╗  ██╔████╔██║█████╗  ██╔████╔██║
  ██║╚██╔╝██║██╔══╝  ██║╚██╔╝██║██╔══╝  ██║╚██╔╝██║
  ██║ ╚═╝ ██║███████╗██║ ╚═╝ ██║███████╗██║ ╚═╝ ██║
  ╚═╝     ╚═╝╚══════╝╚═╝     ╚═╝╚══════╝╚═╝     ╚═╝
  persistent memory for Claude Code

Step 2 — One-sentence pitch:

memem gives Claude Code persistent memory across sessions. It extracts durable lessons (decisions, conventions, bug fixes, preferences) from your completed sessions into markdown files, then surfaces the relevant ones at the start of each new session.

Step 3 — How it works:

  1. You work normally in Claude Code — nothing to do.
  2. Once you opt in via /memem-mine or /memem-mine-history, the Stop hook fires on every conversation turn and spawns a detached mine_delta subprocess.
  3. mine_delta extracts durable memories from the new turns via Claude Haiku and writes them to ~/obsidian-brain/memem/memories/ — memories appear seconds after each turn ends. No background daemon, no 5-minute wait.
  4. Your next session starts with relevant context pre-loaded from memory — no re-explaining the project.

Step 4 — Mining options:

  • /memem-mine — opt in to event-triggered mining (mines new sessions as they happen via the Stop hook)
  • /memem-mine-history — opt in + backfill all past history (uses Haiku API credits)

Or just tell Claude:

  • "start mining new sessions"
  • "start mining everything including history"

If the user is unsure, recommend /memem-mine as the default.

Step 5 — Commands and tools:

Slash commands:

  • /memem — status (full welcome for new users, compact status for returning users)
  • /memem-welcome — this screen (re-show intro)
  • /memem-status — detailed memory system status
  • /memem-doctor — preflight health check with fix instructions
  • /memem-mine — opt in to event-triggered mining (new sessions only)
  • /memem-mine-history — opt in + backfill past history

MCP tools (Claude calls these automatically when useful):

  • memory_save, memory_recall, memory_list, memory_import, transcript_search, context_assemble

Step 6 — Optional Obsidian note:

Memories are plain markdown files at ~/obsidian-brain/memem/memories/. Open that folder as an Obsidian vault for graph view and backlinks. memem works identically without Obsidian.

Step 7 — First action suggestion:

If the user asks "what should I do now?" or seems undecided, answer with one short recommendation:

Start with /memem-mine. It enables memory for new sessions going forward without spending Haiku credits on old history.

Step 8 — Opt-out:

To opt out of event-triggered mining: rm ~/.memem/.miner-opted-in. No daemon to stop — the Stop hook just no-ops when the marker is missing.

Compartilhar X LinkedIn