api2cli-publish-to-npm
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes standard local development commands including
npm whoami,npm view,git remote get-url origin, andbun run build. These are used appropriately to verify the environment, retrieve metadata, and prepare the package artifacts. The commands are procedural and intended for the local developer environment.\n- [PROMPT_INJECTION]: The skill parses project files (package.json,README.md) which could contain untrusted input from a processed repository. \n - Ingestion points: Reads and updates local
package.json,README.md, andSKILL.md.\n - Boundary markers: None used to separate file content from the agent's instructions.\n
- Capability inventory: Shell command execution via
npm,bun, and standard Unix utilities (chmod,mv,cat).\n - Sanitization: No specific instruction filtering is performed on file contents during processing. \n
- Context: The risk is low as the skill facilitates a standard developer workflow and requires explicit user confirmation before any sensitive action like publishing.
Audit Metadata