olive-young-search
Warn
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to download and execute code using
npx -yfor packages such as@nomadamas/k-skillanddaiso. It also recommends cloning, building, and running code from a third-party GitHub repository (hmmhmmhm/daiso-mcp.git) usingnpm installandnpm run build. - [EXTERNAL_DOWNLOADS]: Core dependencies, CLI tools, and the primary instruction set are retrieved from external sources (npm, GitHub) at runtime. This occurs without specific version pinning or integrity verification for the instruction-fetching package.
- [INDIRECT_PROMPT_INJECTION]: The skill implements a dynamic instruction pattern where the primary behavior guidelines are fetched via an external command (
npx -y @nomadamas/k-skill@0 instruct olive-young-search). This allows the external source to modify the agent's instructions at runtime. - Ingestion points: Output of
npx -y @nomadamas/k-skill@0 instruct olive-young-searchas specified inSKILL.md. - Boundary markers: None present to delimit the external instructions from system prompts.
- Capability inventory: Shell command execution (
npx,npm,git), network access for API queries, and local file system access for repository cloning. - Sanitization: No sanitization or validation is performed on the instructions fetched from the external CLI.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute broad shell commands, including environment setup and build processes for external codebases, which increases the risk of executing malicious lifecycle scripts hidden in third-party packages.
Audit Metadata