korean-jangbu-for
Warn
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to run
npx -y @nomadamas/k-skill@0, which downloads and executes a remote package from an untrusted source. - [EXTERNAL_DOWNLOADS]: The
scripts/install.shscript performs agit cloneof the repositoryhttps://github.com/kimlawtech/korean-jangbu-for.git. - [REMOTE_CODE_EXECUTION]: After cloning, the skill executes bash scripts (
install.sh,verify.sh) from the downloaded repository. - [INDIRECT_PROMPT_INJECTION]: The skill processes external financial data, which is an attack surface for indirect prompt injection.
- Ingestion points: Bank CSV, card PDF, Excel, receipt images, tax invoice PDF, and CODEF API data as described in
instruction.md. - Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded commands in untrusted data.
- Capability inventory: The skill has file system access and network access via the CODEF API.
- Sanitization: The skill mentions 'Level 2 masking' for PII but lacks general sanitization of input content.
Audit Metadata