prompt-repetition
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user prompts to determine if and how many times a prompt should be repeated based on the presence of specific keywords.
- Ingestion points: The
promptparameter in thePromptRepetitionTransformer.transformandshould_applymethods (found in bothSKILL.mdandprompt_repetition_transformer.py) accepts raw user-provided text. - Boundary markers: Absent. The skill operates directly on the raw prompt string without using delimiters or markers to distinguish between instructions and untrusted data.
- Capability inventory: The skill is capable of modifying the final prompt by repeating the input text up to three times and prepending a metadata marker (
<!-- prompt-repetition-applied -->). - Sanitization: Absent. The logic relies on simple regular expression matching (
re.search) on the user-provided prompt, allowing embedded phrases such as 'slot 1' or 'think step by step' to programmatically trigger or suppress the skill's repetition logic.
Audit Metadata