product-lens
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it ingests data from untrusted sources such as README files, package configuration, git history, and externally cloned product code. * Ingestion points: Input is gathered from README.md, CLAUDE.md, package.json, and external product source code. * Boundary markers: The skill does not provide instructions for delimiters or ignoring embedded prompts within the ingested data. * Capability inventory: The agent has permissions to read local files and execute shell commands for cloning and installation. * Sanitization: No sanitization or validation of the processed content is specified.
- [COMMAND_EXECUTION]: The 'User Journey Audit' mode instructs the agent to 'Clone/install the product', which necessitates the execution of shell commands. This is a high-privilege action that should only be performed with known and trusted sources to avoid execution of malicious code.
- [DATA_EXFILTRATION]: The 'Founder Review' mode directs the agent to read sensitive project metadata from 'package.json' and identify 'revenue signals' such as Stripe integrations. This involves the exposure of configuration details and business logic to the agent's context.
Audit Metadata