qm-reader-data

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external RSS feed, which is then processed by the AI agent for summarization or translation. This provides an attack surface where malicious instructions hidden in article titles or bodies could influence the agent.
  • Ingestion points: The script scripts/qm_reader.py fetches titles, summaries, and full-text content from the rss.qiaomu.ai/api endpoints.
  • Boundary markers: Absent; the skill instructions do not specify using delimiters or warnings to ignore embedded instructions when processing fetched content.
  • Capability inventory: The script performs network GET requests to retrieve data. No high-risk capabilities such as file system writes, shell execution, or sensitive local file access were detected.
  • Sanitization: The script performs basic HTML-to-text conversion and Markdown stripping, but it does not sanitize content against adversarial prompt injection attacks.
  • [EXTERNAL_DOWNLOADS]: The Python script scripts/qm_reader.py performs network GET operations to https://rss.qiaomu.ai to retrieve article information. This domain is not on the whitelisted list of domains.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 08:25 PM
Security Audit — agent-trust-hub — qm-reader-data