rps-to-buku-ajar

Fail

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to execute local Python scripts via the command line, specifically using the pattern python .codex/skills/scripts/parse_rps.py "<rps_file_or_text>". This method of executing scripts is high-risk when handling untrusted data.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. By interpolating the untrusted rps_file_or_text directly into a shell command, the skill allows an attacker to escape the double-quoted string (e.g., using "; malicious_command; ") and perform arbitrary command execution.
  • [INDIRECT_PROMPT_INJECTION_SURFACE]: The following evidence chain documents the vulnerability surface: 1. Ingestion points: The <rps_file_or_text> placeholder in SKILL.md receives untrusted user content. 2. Boundary markers: Absent; the input is wrapped in quotes but not sanitized or escaped. 3. Capability inventory: The skill has the capability to execute shell commands via the python invocation. 4. Sanitization: Absent; there are no instructions to validate or clean the syllabus text before processing.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 26, 2026, 09:52 AM
Security Audit — agent-trust-hub — rps-to-buku-ajar