NBJ OB1 Agent Memory for OpenClaw

PassAudited by VirusTotal on May 4, 2026.

Overview

Type: OpenClaw Skill Name: nbj-ob1-agent-memory-openclaw Version: 0.1.0 The skill bundle defines a disciplined framework for managing agent memory using the OB1 Agent Memory API. It includes explicit security and privacy instructions in SKILL.md, such as forbidding the storage of secrets, credentials, and private data, and implementing a 'use_policy' to distinguish between human-confirmed instructions and model-generated evidence. No indicators of data exfiltration, malicious execution, or harmful prompt injection were found.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Past task summaries or recalled memories may influence future answers or actions, so incorrect or overbroad memories could carry forward.

Why it was flagged

The skill intentionally recalls and stores persistent operational memory, which can shape future agent behavior.

Skill content
Recall before meaningful work. Write back only compact, provenance-labeled operational memory after the work is complete.
Recommendation

Keep project scoping enabled, review memory before treating it as instruction, and avoid storing secrets, raw transcripts, or private customer data.

What this means

Operational memory may be handled by external systems, which affects privacy, retention, and access controls.

Why it was flagged

The memory workflow depends on external services and an API, so task context and stored summaries may leave the immediate runtime depending on the installed integration.

Skill content
"services": ["Supabase", "OpenRouter"], "tools": ["OpenClaw", "OB1 Agent Memory API"]
Recommendation

Verify the OB1/OpenClaw integration configuration, provider permissions, retention policy, and which data is sent before using it with sensitive work.

What this means

The reviewed skill does not contain executable code, but the real security posture depends on the separately installed memory tools.

Why it was flagged

The package itself is instruction-only and references separate integrations that were not included in the reviewed artifacts.

Skill content
It pairs with the [NBJ OB1 Agent Memory for OpenClaw integration](../../integrations/openclaw-agent-memory/) and the runtime-neutral [OB1 Agent Memory API](../../integrations/agent-memory-api/).
Recommendation

Review and install the referenced OB1/OpenClaw integration only from a trusted source, and check its permissions and credential handling separately.