Ai Safety

"Method over hacks."

TL;DR

Content filters, output guardrails, jailbreak prevention, safety evaluation. [EXPLICIT]

Procedure

Step 1: Discover

• Gather context and requirements

Step 2: Analyze

• Evaluate options per Constitution XIII/XIV

Step 3: Execute

• Implement with evidence tags

Step 4: Validate

• Verify quality criteria met

Quality Criteria

• Evidence tags applied
• Constitution-compliant
• Actionable output

Usage

Example invocations:

• "/ai-safety" — Run the full ai safety workflow
• "ai safety on this project" — Apply to current context

Assumptions & Limits

• Assumes access to project artifacts (code, docs, configs) [EXPLICIT]
• Requires English-language output unless otherwise specified [EXPLICIT]
• Does not replace domain expert judgment for final decisions [EXPLICIT]

Edge Cases

Scenario	Handling
Empty or minimal input	Request clarification before proceeding
Conflicting requirements	Flag conflicts explicitly, propose resolution
Out-of-scope request	Redirect to appropriate skill or escalate