Moses Modes

PassAudited by ClawScan on May 10, 2026.

Overview

This instruction-only skill appears purpose-aligned, but users should understand that it changes agent behavior based on a persistent local mode file and relies on an external governance bundle.

This skill is not showing malicious behavior in the provided artifacts. Before installing, understand that it can influence the agent's behavior across tasks through a local mode file, and review the companion moses-governance bundle if you plan to use the /govern command or any logging/auditing features.

Findings (3)

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.

Low

#ASI01: Agent Goal Hijack

What this means

The agent may behave differently across tasks depending on the selected governance mode.

Why it was flagged

The skill deliberately changes agent behavior globally according to an active mode. This is disclosed and purpose-aligned, but it is broad prompt-level control that users should notice.

Skill content

Load active mode from `~/.openclaw/governance/state.json` before every action. Apply the constraints below as governance guardrails.

Recommendation

Only enable modes you understand, and check the active mode before sensitive or high-impact tasks.

Low

#ASI04: Agentic Supply Chain Vulnerabilities

What this means

The mode-setting behavior may depend on external files or scripts not assessed here.

Why it was flagged

The skill depends on a separate moses-governance bundle and helper script that are not included in the reviewed artifacts, so that dependency's behavior is outside this review.

Skill content

requires:
      - moses-governance
example: |
  # Set mode via operator command: /govern high-security
  # Or directly: python3 ~/.openclaw/workspace/skills/moses-governance/scripts/init_state.py set --mode high-security

Recommendation

Install the referenced moses-governance dependency only from a trusted source and review its scripts before use.

Low

#ASI06: Memory and Context Poisoning

What this means

If logging is implemented by the companion governance bundle, task details or summaries may persist across sessions.

Why it was flagged

The reference modes mention persistent reflection/logging behavior. The provided skill does not implement storage itself, but users should be aware that associated governance tooling may retain context.

Skill content

Reflect on prior interactions. Track what works and what doesn't. Identify capability gaps. Maintain growth log.

Recommendation

Confirm where logs are stored, how to clear them, and avoid placing sensitive information into modes that maintain persistent logs unless you intend that retention.