honcho-memory-integration

What problem does it solve?

Distributed agent teams often suffer from fragmented, noisy, or conflicting memories across projects, users, and sessions, which leads to incorrect retrievals, leaking of context, and loss of important episodic or project-level knowledge. This Skill defines structure and patterns to reliably scope, curate, retrieve, and consolidate persistent memories so agents can maintain continuity and avoid cross-project contamination.

Core Features & Use Cases

• Scoping Hierarchy: Enforce app_id, user_id, and session_id separation to prevent cross-project leakage and enable project-scoped queries.
• Hybrid Recall & Tagging: Combine metadata filtering and semantic similarity for precise retrieval; require minimal tags such as context and status on every memory.
• Lifecycle & Consolidation: Patterns for creating, upserting, archiving, deduplicating, scoring, and scheduled consolidation (cron) to compress episodic traces into semantic memories.
• Conflict Prevention & Sharing: Append-first writes, explicit updates only, deduplication thresholds, and agent-to-agent sharing controls for workspace-level memory coordination.
• Operational Guidance: Troubleshooting steps, secret-handling recommendations, and best practices for async writes and monitoring memory counts.

Quick Start

Consolidate recent session memories into the project-scoped app_id, tag each entry with context and status, and run a deduplication pass to mark superseded items for archival.