v1.3.1

Meow Speech

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 8:12 AM.

Analysis

This is an instruction-only cat-persona skill, with clearly disclosed opt-in scheduling and external-channel check-in guidance that users should enable deliberately.

GuidanceSafe to treat as a persona skill, but only enable proactive check-ins, memory continuity, or Feishu/Telegram delivery if you intentionally want scheduled affectionate messages outside the current chat.

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.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Rogue Agents

SeverityLowConfidenceHighStatusNote

references/scheduler-templates.md

Only schedule when the user has opted in. ... Daily bedtime reminder ... run every day at about 22:00

The skill documents recurring host-side scheduled messages, which can persist after the current interaction, but it also clearly requires opt-in and provides limits.

User impactIf enabled, the assistant may send future check-ins without a new prompt at that moment.

RecommendationOnly enable scheduled check-ins if you want them, review the created jobs, and use the documented opt-out path if they become unwanted.

Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Insecure Inter-Agent Communication

SeverityLowConfidenceHighStatusNote

SKILL.md

It may be used to draft messages for supported delivery surfaces such as Feishu or Telegram. ... only when the user has explicitly enabled that behavior.

The skill can prepare or route messages to external chat channels through the host environment, which changes where messages may appear, but the behavior is disclosed and consent-gated.

User impactOpted-in messages could be delivered outside the local OpenClaw chat, such as to a connected Feishu or Telegram channel.

RecommendationConfirm the exact destination channel before enabling external delivery; use draft-only behavior if you do not want automatic sending.

Memory and Context Poisoning

SeverityLowConfidenceHighStatusNote

references/local-automation-template.md

Store only: proactive_enabled, timezone, preferred_channel, last_user_activity_at, last_proactive_sent_at, pending_idle_job_id, bedtime_job_id

The automation guidance uses persistent host-side state about preferences, channel choice, and activity timing to decide future check-ins.

User impactIf proactive mode is enabled, the system may retain limited scheduling and activity metadata to support future reminders.

RecommendationEnable memory-backed continuity only if desired, and review or clear stored state if you no longer want personalized proactive check-ins.