pharma-daily-report
Warn
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches real-time market data from the Sina Finance public API (hq.sinajs.cn) and requires the official '@larksuite/cli' package from the NPM registry.
- [COMMAND_EXECUTION]: The script 'scripts/daily_pipeline.py' executes the 'lark-cli' binary using 'subprocess.run'. The command construction uses a list of arguments rather than a shell string, which is a secure practice that mitigates common shell injection risks while providing the script control over messaging tools.
- [PROMPT_INJECTION]: The skill includes a file named '.security-scan-passed' which contains unverified claims about the skill's safety and integrity. This constitutes deceptive metadata by attempting to provide false assurance through a self-authoritative safety statement.
- [PROMPT_INJECTION]: Surface for indirect prompt injection detected. Ingestion point: 'scripts/daily_pipeline.py' (fetches stock data from 'hq.sinajs.cn'). Boundary markers: Absent. Capability inventory: 'scripts/daily_pipeline.py' (sends markdown messages to Feishu). Sanitization: Absent. Untrusted external strings (stock names) are interpolated directly into communication templates.
Audit Metadata