sales-thrivecart

Pass

Audited by Gen Agent Trust Hub on Jul 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill implements a persistent learning feature where the agent is instructed to append new insights, workarounds, or tips to the file 'references/learnings.md'. This creates an indirect prompt injection surface because data from untrusted user prompts can be persisted into a file that the agent reads at the start of every session to guide its behavior.
  • Ingestion points: User prompts regarding ThriveCart platform usage and troubleshooting (SKILL.md).
  • Boundary markers: None are present in the instructions to distinguish user-provided tips from system instructions.
  • Capability inventory: File-write access to 'references/learnings.md' is documented as a core feature in SKILL.md and used for context in Step 1.
  • Sanitization: No validation, filtering, or sanitization is performed on the content before it is appended to the reference file.
  • [COMMAND_EXECUTION]: The skill includes a shell command 'npx skills add sales-skills/sales --skill sales-do -a claude-code' in the 'Related skills' section of SKILL.md. This command is used to install additional tools from the same vendor ('sales-skills').
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 19, 2026, 04:16 PM
Security Audit — agent-trust-hub — sales-thrivecart