AI Memory Lite
Personal v0.2: local-first handoff memory for Claude Code, Claude chat, ChatGPT, and Codex in VS Code.
It does not call any API. It reads local transcript files, normalizes them, and creates:
.ai-memory/index.jsonl
.ai-memory/raw/*.jsonl (always gitignored)
.ai-memory/handoff/HANDOFF.md
.ai-memory/handoff/CURRENT_TASK.md
.ai-memory/handoff/DECISIONS.md
.ai-memory/handoff/BUGS.md
.ai-memory/handoff/FILES_TOUCHED.md
.ai-memory/handoff/LAST_SESSION.md
.ai-memory/handoff/NEXT_PROMPT.md
.ai-memory/restored-session/RESUME_PROMPT.md
.ai-memory/restored-session/synthetic-chat.jsonl
AI_CONTEXT.md
CLAUDE.md and AGENTS.md if they do not already exist
What it supports in v0.2
Claude Code
Scans:
~/.claude/projects/**/*.jsonl
or if set:
$CLAUDE_CONFIG_DIR/projects/**/*.jsonl
ChatGPT / Codex
Scans:
~/.codex/history.jsonl
~/.codex/**/*.jsonl
or if set:
$CODEX_HOME/**/*.jsonl
Claude chat history
Two import paths:
- Copy transcript, then run:
AI Memory Lite: Import Clipboard as Claude Chat Transcript
- Export or save Claude chat JSON / JSONL / Markdown / text, then run:
AI Memory Lite: Import Transcript File
Imported Claude chat history is normalized into .ai-memory/index.jsonl and the local Claude normalized JSONL under .ai-memory/raw/.
ChatGPT web or third-party ChatGPT VS Code extensions
Import from clipboard:
AI Memory Lite: Import Clipboard as ChatGPT Transcript
Or import files:
AI Memory Lite: Import Transcript File
Commands
AI Memory Lite: Sync Local History
AI Memory Lite: Generate AI_CONTEXT.md + Handoff
AI Memory Lite: Capture Handoff
AI Memory Lite: Open AI_CONTEXT.md
AI Memory Lite: Open HANDOFF.md
AI Memory Lite: Copy Resume Prompt
AI Memory Lite: Restore Session From Handoff
AI Memory Lite: Import Clipboard as ChatGPT Transcript
AI Memory Lite: Import Clipboard as Claude Chat Transcript
AI Memory Lite: Import Transcript File
Install for local development
- Unzip this folder.
- Open the folder in VS Code.
- Press
F5 to launch an Extension Development Host.
- Open your real coding project in that new window.
- Run the commands above from the Command Palette.
Install as a local VSIX
Install VSCE:
npm install -g @vscode/vsce
Package:
cd ai-memory-lite
vsce package
Install:
code --install-extension ai-memory-lite-0.2.0.vsix
Cross-device workflow
Recommended safe default:
git add AI_CONTEXT.md CLAUDE.md AGENTS.md .ai-memory/index.jsonl .ai-memory/handoff .gitignore
git commit -m "Update AI memory handoff"
git push
On another device:
git pull
Then tell Claude, Codex, or ChatGPT:
Read AI_CONTEXT.md and .ai-memory/handoff/HANDOFF.md, then continue from the latest task.
Or run:
AI Memory Lite: Restore Session From Handoff
That creates and opens:
.ai-memory/restored-session/RESUME_PROMPT.md
.ai-memory/restored-session/SESSION_CONTEXT.md
.ai-memory/restored-session/synthetic-chat.jsonl
The full resume prompt is also copied to your clipboard.
Raw transcripts are always gitignored:
.ai-memory/raw/
.ai-memory/imports/
The extension will keep raw transcripts out of Git automatically. You only need to push the handoff files.
Product direction
The useful thing is not perfect transcript syncing. The useful thing is agent handoff:
- current task
- recent decisions
- active bugs
- files touched
- last session summary
- next resume prompt
That makes the memory usable across Claude, Codex, and ChatGPT even when native histories do not sync cleanly.