dossier-collect
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill builds dossiers by dynamically crawling and exploring external web assets and local files, exposing it to untrusted data payloads.\n
- Ingestion points: External web content retrieved via
WebSearchandWebFetch, alongside local files processed viaRead,Grep, andGlobinSKILL.md.\n - Boundary markers: Absent; there are no defensive delimiters or protective instructions configured to separate or isolate untrusted content from the core execution prompt.\n
- Capability inventory: The system maintains full file system capabilities (
Read,Write), network tools (WebFetch), execution access (Bash), and persistent knowledge base access (mcp__plugin_ruflo-core_ruflo__memory_store).\n - Sanitization: Absent; text extracted from external web targets or files undergoes heuristic regex analysis without sanitization or structural escaping before being integrated into artifact outputs.
Audit Metadata