k-dart
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides Bash and PowerShell scripts to download, extract, and parse corporate registry data from the DART API.
- [EXTERNAL_DOWNLOADS]: Corporate data is fetched from the official South Korean government DART API endpoint (
opendart.fss.or.kr). - [REMOTE_CODE_EXECUTION]: The skill instructs the user to run
npx -y @nomadamas/k-skill@0to retrieve dynamic instructions, which executes code from a package provided by the vendor. - [INDIRECT_PROMPT_INJECTION]: The skill processes XML and JSON data from the external DART API, creating an ingestion surface for potentially untrusted content.
- Ingestion points: XML and JSON responses from
opendart.fss.or.krand theCORPCODE.xmldownload. - Boundary markers: None identified; external data is integrated directly into the agent context.
- Capability inventory: The skill uses network requests (curl), file system access (writing to temporary directories), and shell command execution (grep, awk, unzip).
- Sanitization: The skill uses basic regex-based parsing (awk/grep) but lacks explicit sanitization for adversarial instructions embedded in the financial data.
Audit Metadata