asb-interview-hypotheses
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from an external file to drive its core logic without using boundary markers or sanitization.\n
- Ingestion points: The skill reads business goals from a
GOALS.mdfile (or user-pasted text) during Phase A.\n - Boundary markers: The instructions do not define delimiters or specific instructions to the agent to ignore embedded commands within the input file.\n
- Capability inventory: The skill has the ability to read and write to the local filesystem and invoke other agent skills.\n
- Sanitization: No escaping or validation is performed on the ingested goal text before it is used in the elicitation process.\n- [COMMAND_EXECUTION]: The skill instructions include logic to invoke other agent skills, such as
asb-rude-qaandasb-interview-questions, if they are present in the environment. These are recognized as resources belonging to the same author ('asmartbear') and are used for their intended functional purposes within the interview methodology.
Audit Metadata