setup
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads and installs the 'shinkoku' CLI tool from the author's official repository at
github.com/kazukinagata/shinkokuusing theuvtool manager. This is a standard initialization step for the vendor's application. - [COMMAND_EXECUTION]: The agent executes various shell commands, including
uv tool install,git rev-parse, andshinkoku ledger init, to manage dependencies, verify the environment, and initialize the local SQLite database. - [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface as it reads and processes existing
shinkoku.config.yamland.gitignorefiles from the working directory. While it lacks explicit boundary markers for this data, the risk is minimized by the skill's specific focus on structured tax configuration data. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill handles sensitive personal data (My Number, address, financial status). However, it implements protective measures by saving this data only to local files and ensuring these files are added to
.gitignore. It also provides explicit instructions to keep sensitive identifiers out of tool logs and agent responses.
Audit Metadata