skills/mksglu/context-mode/ctx-stats/Gen Agent Trust Hub

ctx-stats

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The instructions use emphatic keywords ('CRITICAL', 'MUST') and explicit commands ('Do NOT summarize', 'Do NOT collapse') to override the agent's standard behavior of processing and formatting tool outputs.
  • [PROMPT_INJECTION]: By mandating that the entire output of the 'mcp__context-mode__ctx_stats' tool be echoed directly into the response without modification, the skill creates an indirect prompt injection surface. This prevents the agent from sanitizing or filtering any instructions or malicious content that might be present in the tool's output.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 11:26 PM