lifecyclemodel-hybrid-search
Pass
Audited by Gen Agent Trust Hub on Apr 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/run-lifecyclemodel-hybrid-search.mjstriggers shell command execution by invoking a CLI launcher to run thetiangong search lifecyclemodelcommand. - [EXTERNAL_DOWNLOADS]: The skill uses
npm execto fetch the@tiangong-lca/clipackage from the NPM registry during execution. This resource belongs to the skill's author. - [REMOTE_CODE_EXECUTION]: Executing the CLI via
npm execinvolves running remote code downloaded from the NPM registry at runtime. - [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface (Category 8) because it ingests untrusted search queries and filters from user input or JSON files and processes them through a CLI tool with execution capabilities. * Ingestion points: The
queryandfilterfields in request payloads likeassets/example-request.json. * Capability inventory: Subprocess execution viarunTiangongCommandinscripts/run-lifecyclemodel-hybrid-search.mjs. * Boundary markers: No explicit boundary markers or security instructions were identified in the execution scripts to prevent command injection or instruction override. * Sanitization: The wrapper script forwards input arguments directly to the command launcher without explicit sanitization or validation.
Audit Metadata