indykite-authzen-evaluations
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a Bash script (scripts/evaluate-batch.sh) which invokes curl and jq to process and transmit data. This behavior is central to the skill's functionality for interacting with the AuthZEN API.
- [DATA_EXFILTRATION]: The skill transmits API credentials and user tokens to IndyKite's regional endpoints (eu.api.indykite.com and us.api.indykite.com). The implementation includes a security check to ensure that these sensitive tokens are only sent to the authorized vendor hosts.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests JSON-formatted data from local files or standard input to construct API request bodies, creating a surface for indirect prompt injection. Ingestion points: Reads request payload data from the user-specified file or stdin in scripts/evaluate-batch.sh. Boundary markers: No specific delimiters are used to isolate untrusted data within the JSON structure. Capability inventory: Performs network requests using curl and parses data using jq. Sanitization: The script validates that the input is valid JSON using jq before processing, though it does not perform deep schema validation or content sanitization.
Audit Metadata