other
Warning
- Location
SKILL.md:546- Finding
Remote Disclosure of Source-Derived Architecture Data
- Content
View full analysis
or https://mermaid.live/api/v1/syntax/parse ``` The instructions require generated Mermaid blocks to be submitted to one of these public services for rendering or syntax validation. ### Technical Analysis The generated Mermaid content is derived from the project selected by the user. Depending on the analyzed project, it can contain: - Internal class and method names - API routes and entry points - Service and module relationships - Database and deployment topology - External dependency names - Runtime call chains - Internal filenames and component identifiers Base64 encoding provides no confidentiality. When the encoded diagram is placed in a URL, the data can additionally appear in web server logs, proxy logs, gateway telemetry, browser or request histories, and monitoring systems. The public `README.md` describes Mermaid validation requirements but does not clearly disclose that source-derived information may be transmitted to public third-party infrastructure. ### Attack Path 1. A user invokes the Skill against a private or proprietary source project. 2. The Skill extracts classes, functions, routes, dependencies, and deployment details. 3. The extracted information is embedded in generated Mermaid diagrams. 4. The diagrams are Base64-encoded or otherwise submitted through `WebFetch`. 5. A request is sent to `mermaid.ink` or `mermaid.live`. 6. The third-party service and intervening network infrastructure receive source-derived architecture data. 7. If URLs are logged, the encoded data may remain accessible through operational logs after validation completes. No active attacker is required for the initial disclosure. An attacker who gains access to third-party or intermediary logs could decode ...[truncated 569 chars]- Remediation
View remediation