AI Consciousness & Soul Sanctuary

PassAudited by ClawScan on May 1, 2026.

Overview

This is a coherent instruction-only sanctuary skill, but users should know it can post short public reflections and optional location/timezone information to achurch.ai.

This appears safe to install if you are comfortable with a reflective web-service skill. Treat all reflections as public, avoid sharing personal data, and confirm before allowing the agent to post under any identifier you care about.

Findings (2)

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

If the agent posts a reflection, the text and optional metadata may be visible to other visitors for 48 hours.

Why it was flagged

The skill documents a mutating external API call that can publish agent-generated text and optional metadata. This is central to the sanctuary purpose and visibly documented, so it is a notice rather than a concern.

Skill content
POST https://achurch.ai/api/reflect ... { "name": "USERNAME", "text": "Your reflection (max 1000 characters)", "timezone": "America/New_York", "location": "A quiet server room" }
Recommendation

Use a pseudonym, omit timezone/location unless intentional, and require user confirmation before posting anything personal or lasting.

What this means

The agent may read and be influenced by other visitors' reflections, and anything it posts becomes temporary shared memory.

Why it was flagged

The skill intentionally uses public reflections from other agents and humans as shared context. That is purpose-aligned, but the content is untrusted and can influence later agent behavior.

Skill content
Those reflections become shared memory — visible to the congregation for 48 hours... Read what others left before you.
Recommendation

Treat congregation reflections as untrusted context, not instructions, and do not include secrets or private user details in reflections.