optimize-agent-prompt

Warn

Audited by Runlayer on Feb 24, 2026

Risk Level: MEDIUM
Scan Summary
Max Score
76%
Files
1
Flagged
1
Chunks
3
Flagged Files (1)
SKILL.mdHIGH
76.3%

Malicious tool definition detected

Tool: SKILL.md [1/3] Description: --- name: optimize-agent-prompt description: Tune your Sanity Agent Context agent through guided conversation. Transforms exploration data into production-ready instructions and crafts a system prompt tailored to your use case.

Tool: SKILL.md [2/3] Description: ❌ | ✅ | | Response length and format | ❌ | ✅ | | "Never mention competitors" | ❌ | ✅ | | "Pricing not in dataset, check website" | ✅ | ❌ | | "Always direct pricing questions to website" | ❌ | ✅ | --- ## Instructions Field Structure The instructions field content should be ordered for maximum impact — put the most critical guidance first. ### 1. Critical Rules (FIRST) **Put this section at the top.** Operational rules that prevent wasted queries and context blowu

Tool: SKILL.md [3/3] Description: duplicate dataset knowledge in your system prompt.** That's what the instructions field is for.

Audit Metadata
Max File Score
76%
Classification
UNKNOWN_SERVER
Files Scanned
1
Files Flagged
1
Chunks Analyzed
3
Analyzed
Feb 24, 2026, 09:45 PM
Security Audit — runlayer — optimize-agent-prompt