Skip to content

Commit e84f9e1

Browse files
sedelevclaude
andcommitted
Reposition feed-memex as legacy; point users at memex-sync scan-claude
The 'Live MCP memory — feed-memex' section in both README's framed this command as the canonical way to feed memex from claude-backup. That made claude-backup a de-facto memex prerequisite — install Python, install claude-backup, then run feed-memex to bootstrap memex's corpus. memex shipped a self-contained ingester yesterday (memex-sync scan-claude). It walks the same ~/.claude/projects and Cowork session trees, parses the same JSONL shape, and writes byte-identical inbox files. No Python required. Updated both README's: • Top of the feed-memex section now leads with a heads-up callout: "if your only goal is to feed memex, use `npx memex-sync scan-claude` — you don't need claude-backup at all" • feed-memex remains documented (the command still works; people with existing crons or scripts shouldn't break) but is explicitly marked legacy / dual-use • Added the byte-compatibility note so users know they can switch between tools without re-importing This realigns the two projects with their proper roles. claude-backup is the standalone Markdown export utility — Plan B for users who want their conversations as readable .md files for backup, sharing, or reading offline. memex is the AI-memory layer that ingests conversations into a searchable index. They were always meant to be independent; the README's just hadn't caught up. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 8c7a95a commit e84f9e1

2 files changed

Lines changed: 20 additions & 4 deletions

File tree

README.md

Lines changed: 10 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -223,7 +223,15 @@ Workflow:
223223

224224
A 200-message session typically packs into 80–200 KB of text — fine for context windows of any modern hosted assistant.
225225

226-
### Live MCP memory — `feed-memex`
226+
### Live MCP memory — `feed-memex` (legacy)
227+
228+
> **Heads-up:** if your only goal is to feed [memex-mvp](https://github.com/parallelclaw/memex-mvp), you don't need claude-backup at all. Memex now ships a self-contained ingester:
229+
>
230+
> ```bash
231+
> npx memex-sync scan-claude # imports the same Code + Cowork history, no Python needed
232+
> ```
233+
>
234+
> Use `claude-backup feed-memex` only when you already have claude-backup installed for Markdown export and want a single tool to do both jobs.
227235
228236
If you run [memex-mvp](https://github.com/parallelclaw/memex-mvp) (a separate local MCP server), `claude-backup feed-memex` writes a clean dialogue-only JSONL of every session into memex's inbox folder (`~/.memex/inbox/`). Memex picks them up via `chokidar`, indexes via SQLite + FTS5, and exposes them through MCP to **any compatible AI agent** — Cursor, Cline, Claude Code, Continue, Zed.
229237
@@ -232,7 +240,7 @@ claude-backup feed-memex # write all sessions to ~/.memex/inbox/
232240
claude-backup feed-memex --dry-run # show what would be written
233241
```
234242
235-
Output is idempotent — re-run anytime, memex dedupes by stable msg_id. Once set up, your Cursor agent can just `memex_search("the migration we discussed in April")` and surface real results from your past Code/Cowork conversations. **Zero paste.** Zero context-switching.
243+
Output is idempotent — re-run anytime, memex dedupes by stable msg_id. Output format is byte-compatible with `memex-sync scan-claude`, so you can switch between the two without re-importing.
236244
237245
Filename convention: `code-<8char>.jsonl` for Claude Code sessions, `cowork-<8char>.jsonl` for Cowork. Memex distinguishes the two via the prefix and tags them with separate `source` values, so you can filter `memex_search` by source.
238246

README.ru.md

Lines changed: 10 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -223,7 +223,15 @@ Workflow:
223223

224224
Сессия из 200 сообщений обычно укладывается в 80–200 КБ текста — нормально для context window любого современного хостед-ассистента.
225225

226-
### Живая MCP-память — `feed-memex`
226+
### Живая MCP-память — `feed-memex` (legacy)
227+
228+
> **Важно:** если ваша единственная цель — кормить [memex-mvp](https://github.com/parallelclaw/memex-mvp), claude-backup вам не нужен. Memex теперь самодостаточен:
229+
>
230+
> ```bash
231+
> npx memex-sync scan-claude # импортит ту же Code + Cowork историю, без Python
232+
> ```
233+
>
234+
> Используйте `claude-backup feed-memex` только если у вас уже стоит claude-backup для Markdown-экспорта и хочется чтобы один tool делал и то и другое.
227235
228236
Если у вас установлен [memex-mvp](https://github.com/parallelclaw/memex-mvp) (отдельный локальный MCP-сервер), команда `claude-backup feed-memex` пишет clean dialogue-only JSONL каждой сессии в memex'овский inbox (`~/.memex/inbox/`). Memex подхватывает через `chokidar`, индексирует в SQLite + FTS5, и даёт доступ через MCP **любому совместимому AI-агенту** — Cursor, Cline, Claude Code, Continue, Zed.
229237
@@ -232,7 +240,7 @@ claude-backup feed-memex # записать все сессии в ~/
232240
claude-backup feed-memex --dry-run # показать что будет записано
233241
```
234242
235-
Вывод идемпотентный — запускайте сколько угодно, memex дедуплицирует по стабильным msg_id. После настройки ваш Cursor-агент может вызвать `memex_search("миграция о которой мы говорили в апреле")` и получить реальные результаты из прошлых Code/Cowork разговоров. **Никакой пасты.** Никаких переключений контекста.
243+
Вывод идемпотентный и **байт-совместимый с `memex-sync scan-claude`** — можно переключаться между tools без переимпорта. Memex дедуплицирует по стабильным msg_id.
236244
237245
Конвенция имён: `code-<8char>.jsonl` для Claude Code сессий, `cowork-<8char>.jsonl` для Cowork. Memex различает их по префиксу и помечает разными `source`-значениями — можно фильтровать `memex_search` по источнику.
238246

0 commit comments

Comments
 (0)