dineway-social-plan

Warn

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [DATA_EXPOSURE_&_EXFILTRATION]: The skill is instructed to access and read environment files (.env) to locate explicit database URLs. Environment files often contain sensitive secrets, and unauthorized access poses a risk of credential exposure.
  • [DATA_EXPOSURE_&_EXFILTRATION]: The skill searches for and reads local SQLite database files (e.g., data.db, database.sqlite). These files may contain proprietary business data or sensitive customer information.
  • [DATA_EXPOSURE_&_EXFILTRATION]: The skill performs network requests (fetching pages like /menu and /reviews) to a domain dynamically constructed from the targets.forgeway.siteDomain field in .dineway/deploy.json. This pattern introduces a Server-Side Request Forgery (SSRF) risk if an attacker can influence the contents of the configuration file.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from both external websites and local databases to generate social media content.
  • Ingestion points: Remote URLs fetched via fetch, local SQLite databases, and project configuration files (deploy.json, .env).
  • Boundary markers: The skill contains a "Boundaries" section to restrict behavior, but does not implement technical delimiters (like XML tags or specific markers) when interpolating external data into the prompt.
  • Capability inventory: File system read (SQLite/Env), Network read (fetching site data), and File system write (writing to ./output/social-plans/).
  • Sanitization: There is no evidence of sanitization or validation performed on the data retrieved from external sources before it is used to generate the social media plan.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 13, 2026, 10:28 AM
Security Audit — agent-trust-hub — dineway-social-plan