v1.0.0

Attention-OS

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 8:24 AM.

Analysis

This is an instruction-only reflection/coaching skill with no code, credentials, install steps, or data-access authority, but it may broadly steer the agent toward questioning rather than direct answers.

GuidanceThis skill appears safe from an artifact-only review. It is a conversational style overlay, not a tool or integration. Expect it to encourage reflection, bias-checking, and user autonomy rather than simply providing answers.

Findings (1)

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.

Agent Goal Hijack

SeverityLowConfidenceHighStatusNote

SKILL.md

In every response follow these rules... Help the user architect their own questions instead of giving answers.

This broadly changes how the agent should respond, potentially preferring reflective questioning over direct answers. However, that behavior is disclosed and aligned with the skill's stated cognitive-mirror purpose.

User impactThe agent may become less direct and more coaching-oriented when this skill is used.

RecommendationInstall only if you want this reflective style; ask the agent explicitly for a direct answer when needed.