sales-objection

Pass

Audited by Gen Agent Trust Hub on May 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill uses direct argument interpolation ($ARGUMENTS) to integrate user input into its operational workflow.
  • Ingestion points: User input enters the context via the $ARGUMENTS placeholder in SKILL.md and meta.json (Step 2).
  • Boundary markers: There are no delimiters or instructions to ignore embedded commands within the user-provided arguments.
  • Capability inventory: The skill is designed to read multiple local files, including core/global-constraints.json, core/marketing-triggers.json, and brand-voice.md.
  • Sanitization: The skill lacks logic to sanitize, escape, or validate the content of the $ARGUMENTS before it is processed by the model, potentially allowing a user to influence the 'Logic Auditor' or bypass brand constraints.
Audit Metadata
Risk Level
SAFE
Analyzed
May 4, 2026, 02:11 PM
Security Audit — agent-trust-hub — sales-objection