OpenClaw integration, minus the hand-waving

If you already use OpenClaw, Memory Crystal should behave like native infrastructure: invisible when healthy, obvious when broken.

Integration model

Memory Crystal currently integrates with OpenClaw in two layers:

1. Plugin hooks in plugin/index.js capture turns, trigger wake briefings, surface preflight warnings, and register the remote memory tools.
2. Context-engine behavior handles local summaries, compaction, and prompt-aware summary injection inside the running session.

That means you don’t manually serialize context. The plugin keeps capture, recall, and local-session context management running in the background.

Install the current plugin

curl -fsSL https://memorycrystal.ai/crystal | bash

The installer downloads the current plugin files, updates the OpenClaw memory slot, and walks you through browser auth or API-key entry. To update an existing install:

curl -fsSL https://memorycrystal.ai/crystal/update | bash

Wake briefings explained

Wake briefing is not just “recent chat.” It combines:

• high-salience long-term memories
• recent message matches
• open goals and recent decisions

This gives the model a compact operational snapshot before any user prompt.

Recall flow under the hood

When you ask something memory-relevant, the plugin:

1. builds a semantic query from current user intent
2. searches long-term memories plus recent messages
3. layers in graph-backed context for ownership, dependency, and connection queries
4. injects the top-ranked slice as context snippets

The model never sees the entire database. It sees the top-ranked slice.

Hook reliability checklist

• log hook registration at startup
• fail closed on invalid key (don’t pretend memory works)
• idempotent install scripts for upgrades
• structured errors for 401/429/500
• dynamic version detection in the installer instead of hardcoded release strings

Debugging empty recall

Run a controlled test:

1. Save deterministic fact: “Project codename is Aurora-7.”
2. Start new session.
3. Ask: “What is project codename?”

If empty, check API key, channel scope, and whether capture events fired.

Production tuning

• Keep maxRecallItems small (6–10)
• Use category-aware recall modes (decision/project/people)
• checkpoint after releases and incident retros

OpenClaw performs best when memory context is sharp, not bloated.

Bottom line

Integration is straightforward. The hard part is memory hygiene and retrieval quality. Get hooks stable first, then optimize ranking.