recoup-internal-sales

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill recommends that the operator execute npx skills add to install the unslop plugin from https://www.skills.sh/cursor/plugins/unslop. This command downloads and executes code from an external domain not recognized as a trusted vendor in the internal list.
  • [INDIRECT_PROMPT_INJECTION]: The skill's workflow for generating sales pitches and meeting prep involves ingesting raw customer chat data, which creates a potential surface for indirect prompt injection.
  • Ingestion points: The instructions in SKILL.md and references/meeting-prep.md direct the agent to read chat_messages.parts (customer chat bodies) from the Supabase database and Recoup API.
  • Boundary markers: The instructions lack explicit delimiting or specific instructions to the agent to treat chat content as data only and to ignore any embedded instructions.
  • Capability inventory: The skill has extensive capabilities, including write access to the Attio CRM (for logging follow-ups and advancing leads) and read access to Stripe, Privy, and Supabase financial/user data.
  • Sanitization: No sanitization or filtering of the raw chat bodies is described before the content is used to inform the agent's synthesis and pitch generation.
  • [EXTERNAL_DOWNLOADS]: The skill points to an external source (https://www.skills.sh/cursor/plugins/unslop) for a recommended plugin, introducing a dependency on unverified external infrastructure.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 07:05 PM
Security Audit — agent-trust-hub — recoup-internal-sales