Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Version 1.0.4 of auto-drive - Updated tags
What This Skill Does
Stores agent memories as a permanent linked list on the Autonomys decentralized storage network. Each entry gets a CID and points to the previous one, forming a chain. A new agent instance can reconstruct its full history from a single head CID.
Unlike local files or cloud storage, the memory chain survives total environment loss and is recoverable on any machine from a single CID with no credentials beyond the API key.
When to use it
- Saving an agent decision before ending a session
- Checkpointing a long research or coding session
- Uploading a project file to immutable decentralized storage
- Rebuilding agent identity after total local state loss
- Sharing a permanent, content-addressed file snapshot via CID
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Save a memory that we decided to use TypeScript for this project
- 1Wraps the decision text as a JSON memory payload with type and content fields
- 2Reads the previous head CID from the local state file
- 3Runs autodrive-save-memory.sh with the payload and agent name
- 4Uploads the new entry to Auto Drive, receiving a new CID
- 5Updates the state file and pins the new head CID to MEMORY.md
Returns the new CID, previous CID, and current chain length (e.g. {"cid": "bafk...xyz", "previousCid": "bafk...def", "chainLength": 6})
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.