← back to partnerships-architect

SkillSpector · partnerships-architect

independent scanner by NVIDIA · skill by Alireza Rezvani · how it works ↗

PASSmax severity: LOWrisk score: 10

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance …

scanned 2026-07-22

Findings (3)

MEDIUMMemory Poisoningconfidence: 0.8

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

scripts/revshare_modeler.py

MEDIUMMemory Poisoningconfidence: 0.8

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

scripts/revshare_modeler.py

MEDIUMMemory Poisoningconfidence: 0.8

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

scripts/revshare_modeler.py

What the verdicts mean

SkillSpector reports on SkillMD's shared five-tier scale. See how SkillSpector works ↗.

PASSthis skill

Overall severity LOW (risk score in the safe range)

CAUTION

Overall severity MEDIUM

WARNING

Overall severity HIGH

FAIL

Overall severity CRITICAL

INCONCLUSIVE

Scan could not complete