framing-decisions
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided inputs for decision actions, states, and utility values. This data is written to a JSON file and analyzed by a local script, creating a potential surface for indirect prompt injection if the user provides malicious content within the decision fields.
- Ingestion points: User-provided decision parameters defined in the workflow and template sections of SKILL.md.
- Boundary markers: No specific instructions are provided to the agent to treat this data as untrusted or to ignore embedded instructions within the decision fields.
- Capability inventory: The skill invokes a local script using
node scripts/calc.jsto process the generated data file. - Sanitization: No explicit sanitization or validation logic is mentioned in the provided markdown instructions.
- [COMMAND_EXECUTION]: The skill executes a local Node.js script (
scripts/calc.js) to perform utility calculations on user-provided data. This is the primary function of the skill and uses internal skill components.
Audit Metadata