title: "OB1 Agent Memory for OpenClaw: Reference Links"

type: "guide"
label: "Guide"
project: "OpenClaw Crossed The Runtime Line"

OB1 Agent Memory for OpenClaw: Reference Links

This is a reference index for the OB1/OpenBrain Agent Memory release. It is not the full implementation guide. The full guide, contracts, examples, plugin package, skill, and supporting assets live in the OB1 repo and on ClawHub.

Start Here

• OB1 GitHub repo
• Official ClawHub package listing
• OpenClaw Agent Memory recipe
• OpenClaw integration package
• OpenClaw native plugin source
• OpenClaw Agent Memory skill
• Nate's Substack
• NateBJones.com

What This Release Includes

OB1 Agent Memory gives OpenClaw workflows a governed memory layer:

• Recall relevant context before meaningful work starts.
• Write back compact operational memory after work finishes.
• Preserve source, model, runtime, task, confidence, scope, and confirmation metadata.
• Keep inferred/generated memory as evidence until a human confirms it.
• Give builders recipes for code review, TaskFlow handoffs, and provenance-aware memory.

Core Recipe

• Recipe README
  The main implementation guide for using OpenBrain/OB1 as a durable memory layer for OpenClaw workflows.

• Recipe metadata
  Machine-readable package metadata for the recipe.

Contracts

• Recall request schema
  Defines what an OpenClaw workflow sends to OB1 before starting work.

• Recall response schema
  Defines how OB1 returns scoped memories, provenance, freshness, and use policy.

• Write-back schema
  Defines how OpenClaw writes back decisions, outputs, lessons, unresolved questions, artifacts, and metadata.

Example Payloads

• Example recall request
• Example write-back request
• Example usage report

Companion Recipes

• Code Review Memory recipe
  Stores reusable repo lessons from PR reviews: conventions, recurring bugs, risky files, tests, maintainer preferences, and false positives.

• Code Review recall example

• Code Review write-back example

• TaskFlow Work Log recipe
  Lets long-running OpenClaw work leave behind compact handoff memory: what was attempted, what changed, what failed, blockers, and next-agent context.

• TaskFlow recall example

• TaskFlow write-back example

Trust, Safety, And Provenance

• Safe Agent Memory and Provenance guide
  Explains why inferred/generated memory should not silently become instruction, and how OB1 handles confirmation, evidence, stale memory, disputed memory, and trusted imports.

• Agent Memory portability notes
  Notes on keeping OB1 Agent Memory runtime-neutral instead of over-coupled to OpenClaw.

Database Schema

• Agent Memory schema README
  Overview of the database model for governed agent memory.

• Agent Memory SQL schema
  Sidecar schema for memory records, provenance, use policy, review state, source references, recall traces, trace items, relations, and audit events.

Runtime-Neutral API

• Agent Memory API integration
  Runtime-neutral API surface for recall, write-back, usage reporting, review actions, inspector reads, recall traces, and health checks.

• API README

• API implementation

• Live smoke test

• Demo seed script

OpenClaw Plugin

• OpenClaw integration folder
  Contains the native OpenClaw plugin package, publishing notes, release notes, and plugin skill bundle.

• Integration README

• ClawHub publishing notes

• Plugin README

• Plugin manifest

• Plugin package.json

• Plugin source

• Native OpenClaw smoke test

OpenClaw Skill

• Skill README
• Skill instructions
• Skill metadata

The skill tells OpenClaw agents how to behave around memory: recall before meaningful work, write back compact operational memory, avoid raw transcript dumps, preserve provenance, and never treat inferred memory as confirmed instruction.

Visual Assets

• Agent Memory visual asset index
• Diagram source files
• Exported SVG diagrams
• Video-ready PNG diagrams
• Dashboard screenshots
• Promotional assets
• OB1/NBJ brand assets

Dashboard Walkthrough Assets

• Dashboard walkthrough folder
• Walkthrough README

This folder contains the screenshot capture scripts, demo seed data, guide generation scripts, and Remotion-based video walkthrough source for the OB1 Agent Dashboard.

Release Notes

• OpenClaw Agent Memory v0.1.0 release notes
• OpenClaw Agent Memory v0.1.1 release notes

Suggested Reader Path

1. Start with Nate’s Substack article for the strategy and framing.
2. Use the OpenClaw Agent Memory recipe for the implementation pattern.
3. Review the recall and write-back contracts.
4. Install or inspect the official ClawHub package.
5. Use the companion recipes for concrete workflows:
   • Code Review Memory
   • TaskFlow Work Log
6. Read the Safe Agent Memory and Provenance guide before allowing agent-written memory to become instruction-grade.