Skills
skills/tiangong-ai/skills/ai-tech-summary/Gen Agent Trust Hub

ai-tech-summary

Pass

Audited by Gen Agent Trust Hub on Feb 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted RSS content, creating an indirect prompt injection surface where embedded instructions in the feed could influence agent behavior.
  • Ingestion points: scripts/time_report.py extracts text from database fields including title, summary, and fulltext_text (extracted from entries and entry_content tables).
  • Boundary markers: The skill uses a JSON data structure to provide context to the agent, but the instructions in SKILL.md lack explicit directives to ignore instructions that may be contained within the records.
  • Capability inventory: The tool executes a Python script that reads from a local SQLite database and can write a JSON context file to a path specified by the --output parameter.
  • Sanitization: The script performs whitespace normalization and text truncation but does not implement specific sanitization to filter out malicious instructions or bypass attempts within the retrieved RSS data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 24, 2026, 03:03 PM