html-to-generateblocks
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process user-supplied HTML and CSS, which provides an attack surface for instructions hidden in code comments or attributes.
- Ingestion points: Untrusted source HTML and CSS are inventoried and mapped into block structures in
SKILL.mdandreferences/conversion-patterns.md. - Boundary markers: The instructions lack specific delimiters or warnings to the agent to disregard instructions that might be embedded within the source DOM or CSS.
- Capability inventory: The skill writes output to the local file system and invokes a local Python script (
preflight.py). - Sanitization: There are no instructions for sanitizing or filtering the source data for malicious natural language prompts before processing.
- [COMMAND_EXECUTION]: The skill requires the execution of a local utility script to process the generated files.
- Evidence: The 'Delivery' section in
SKILL.mdinstructs the agent to runpreflight.py FILE --id-scope SCOPE.
Audit Metadata