Back to skill
Skillv1.0.3
ClawScan security
Wayfound · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 12, 2026, 6:48 PM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's requests and instructions align with its description: it only edits your agent's SOUL.md and memory files and suggests adding a daily cron job; it asks for no credentials, installs, or external downloads.
- Guidance
- This skill appears coherent and low-risk, but before installing: confirm you want the agent to (1) edit your SOUL.md permanently, (2) create daily review files in memory/, and (3) schedule an OpenClaw cron job. The SKILL.md instructs the agent to require your explicit approval before creating the cron — keep that requirement. Decide how (and whether) you want alerts delivered externally, back up SOUL.md if it’s important, and periodically review recent review files to ensure the self-assessments are useful and not drifting.
Review Dimensions
- Purpose & Capability
- okName/description (self-supervision via rubric + daily review) match the instructions: add rubric to SOUL.md, run a daily openclaw cron job that reads memory and writes a short review. The requested actions (read/write SOUL.md and memory files, schedule a cron) are appropriate for that purpose and no unrelated permissions or credentials are requested.
- Instruction Scope
- okSKILL.md stays on-topic: it tells the agent to add a rubric to SOUL.md, schedule a cron job using the OpenClaw CLI, read today's memory file, write memory/review-YYYY-MM-DD.md, and surface issues. These are scoped to the agent's identity and memory. Note: the instructions do modify persistent files (SOUL.md, memory/*) and create a scheduled job, which is expected for this feature and is explicitly gated by 'require your user's explicit approval before setting this up.'
- Install Mechanism
- okInstruction-only skill with no install spec and no code files — minimal disk/write footprint beyond the explicitly described edits to SOUL.md and memory files. Low install risk.
- Credentials
- okNo environment variables, credentials, or external config paths are requested. Optional alert delivery to external channels is mentioned but left to the user's configuration; nothing is hard-coded or required.
- Persistence & Privilege
- noteThe skill creates persistent artifacts (rubric section in SOUL.md, daily review files in memory/) and schedules a cron job via the OpenClaw CLI. always: false (not force-included) and the SKILL.md instructs to get explicit user approval before scheduling the cron — this keeps privilege reasonable. Users should be aware the agent will be granted permission to write those files and run the scheduled job if they approve.