dependency-upgrade
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches documentation and migration codemods from well-known sources, such as the official React repository on GitHub.
- [REMOTE_CODE_EXECUTION]: Documents the use of
jscodeshiftto execute transformation scripts from remote URLs. The provided example uses a trusted source from thereactjsorganization. - [COMMAND_EXECUTION]: Utilizes various package management commands (npm, yarn) and analysis tools (npm-check-updates, madge) to manage dependencies and visualize project structures.
- [DYNAMIC_EXECUTION]: Includes a custom migration script that dynamically reads, modifies, and writes back to local source files using
fsandglobmodules. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources like
CHANGELOG.mdand parses local project files, which could potentially expose the agent to instructions embedded in that data. - Ingestion points: Remote changelogs via curl, local project files via glob search.
- Boundary markers: None present in the example processing scripts.
- Capability inventory: Shell command execution (npm/yarn), remote code execution (jscodeshift), and local file system writes.
- Sanitization: No sanitization or validation of the ingested content is demonstrated in the provided scripts.
Audit Metadata