llm-streaming-response-handler

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's documentation and scripts reference well-known, industry-standard packages such as tiktoken, node-fetch, and the Vercel AI SDK (ai). These are established libraries within the AI development ecosystem.\n- [COMMAND_EXECUTION]: Provides local developer tools stream_tester.ts and token_counter.ts. These scripts are intended to help developers verify streaming endpoints and analyze text data during the development process.\n- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection within its local testing utility.\n
  • Ingestion points: scripts/stream_tester.ts reads untrusted data from network streams using the response.body.getReader() method.\n
  • Boundary markers: The script does not implement boundary markers or instructions to ignore embedded commands within the incoming data stream.\n
  • Capability inventory: The skill is configured with access to the Bash tool, representing a potential escalation path if malicious instructions from a stream are processed.\n
  • Sanitization: Incoming stream content is written directly to the terminal's standard output without sanitization or escaping of potentially malicious terminal escape sequences.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 01:52 PM