xmpp-enumeration
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses Python socket and ssl modules, along with nmap, to actively interact with and probe XMPP services on ports 5222, 5223, and 5269.\n- [DATA_EXFILTRATION]: Discovered usernames and Multi-User Chat (MUC) room history are collected and saved to the engagement/evidence/ directory for persistence and analysis.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from external XMPP servers (room history and search results), creating an indirect prompt injection surface.\n
- Ingestion points: socket.recv() calls in Step 3b (authenticated user search) and Step 4 (MUC room history retrieval).\n
- Boundary markers: None present in the provided scripts or instructions to separate untrusted data from agent context.\n
- Capability inventory: File writing via the Write tool and shell command execution via python3 and nmap.\n
- Sanitization: Content is decoded with errors="replace" but lacks sanitization or filtering of embedded instructions before being processed by the agent.
Audit Metadata