B
85/100 First Seen
Feb 18, 2026
Last Scanned
Feb 20, 2026
Findings
1
Score
85/100
HIGH 1
Findings (1)
HIGH
Self-modifying agent instructions
L11 Detects skills that write or promote content into agent instruction files
write to CLAUDE.md FIX
Remove or restrict the skill's ability to modify CLAUDE.md or agent configuration files. Self-modifying agent configurations can be exploited for persistent attacks.
FP?
Likely FP if the skill is a project management tool where updating CLAUDE.md is an intended workflow feature (e.g., appending project notes), though the risk remains.