chatgpt-import_skill

This skill helps you migrate and index ChatGPT conversations into OpenClaw memory for fast, searchable past chats.

Python

2.5k

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill chatgpt-import

_meta.json286 B
SKILL.md1.7 KB

Overview

This skill imports ChatGPT conversation history into OpenClaw so past chats become searchable via OpenClaw's memory search. It converts ChatGPT exports to markdown, creates embeddings, and stores them in a local SQLite memory database. Use it to migrate histories, build searchable archives, or give OpenClaw access to legacy conversations.

How this skill works

The workflow converts a ChatGPT export (conversations.json) into markdown conversation files, then calls an embedding service to generate vector representations. Those embeddings are batched and stored in a SQLite database designed for OpenClaw memory search. Finally, you point OpenClaw to the generated database so the gateway can query your imported history.

When to use it

Migrating conversation history from ChatGPT to OpenClaw
Building a searchable archive of past chats for reference or compliance
Providing OpenClaw access to legacy conversations for context-aware agents
Consolidating multiple chat exports into one memory database
Filtering and embedding large volumes of historical chats for semantic search

Best practices

Export directly from ChatGPT using the official export flow before running import scripts
Run the conversion step with --min-messages to skip trivial or boilerplate chats
Scan and remove any secrets from conversations before embedding; consider automated redaction
Use a scoped OpenAI API key and monitor embedding usage to control costs
Back up original exports and generated databases before re-running or modifying them

Example use cases

Onboard a new OpenClaw agent with historical ChatGPT context for improved responses
Create a searchable institutional knowledge base of team conversations and decisions
Preserve customer support chats from ChatGPT and enable fast retrieval by semantic queries
Audit past chats for policy or compliance reviews with full-text and semantic search
Merge multiple ChatGPT exports into a single searchable memory store for team access

FAQ

Yes. The embedding step sends text to the embedding API and requires an API key. Use a scoped key if you need to limit access.

Will my API key be stored in the database?

No. The generated SQLite memory database does not store your API key. The key is only used at runtime to request embeddings.

Can I filter or redact content before embedding?

Yes. Run a preprocessing pass on the converted markdown to remove or replace sensitive data before calling the embed script.