openclawdy_skill

Overview

This skill provides a persistent memory infrastructure for autonomous agents, offering semantic recall, reputation tracking, cross-agent pools, and time-travel snapshots. It uses wallet-based signing for authentication so agents retain private key control while the platform verifies signed requests. Data is stored in managed vector DB and relational storage with TLS and encryption at rest.

How this skill works

Agents sign a timestamped message with their wallet and include the address, signature, and timestamp in request headers. The service verifies signatures per request and stores agent-specific vaults that persist until deleted. Core APIs let agents store memories, run semantic recall, manage reputation scores, share into pools, and create or restore memory snapshots.

When to use it

• Give an agent long-term memory for facts, preferences, decisions, and learnings across sessions.
• Share knowledge among a team of agents using cross-agent memory pools.
• Track which memories lead to successful outcomes with reputation-scored memories.
• Debug agent behavior by snapshotting and restoring memory states before and after changes.
• Export or permanently delete an agent vault for backup and data governance.

Best practices

• Sign only the required timestamp message from the wallet; never expose private keys to the service.
• Classify memories by type (fact, preference, decision, learning, history, context) to improve recall relevance.
• Use reputation updates to surface strategies that have proven effective over time.
• Create snapshots before major updates or experiments and use compare to audit changes.
• Limit semantic recalls with sensible query and limit parameters to stay within rate limits.

Example use cases

• A personal assistant agent stores user preferences and recalls them to personalize responses.
• A swarm of research agents shares bug reports and fixes in a shared pool for collective troubleshooting.
• A trading agent snapshots its memory state before a strategy change to compare outcomes later.
• A support bot records successful solutions with high reputation so future recalls surface proven fixes.
• An operations agent exports its vault for archival or migrates memories between environments.