testing
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to discover and execute commands from repository-controlled files such as
package.json, CI configurations, and contributor documentation. This creates a surface where an attacker who can modify these repository files could inject malicious instructions or shell commands that the agent would then execute. 1. Ingestion points:package.json, repository scripts, CI configuration, and contributor documentation as specified inSKILL.mdandreferences/typescript.md. 2. Boundary markers: The instructions lack specific boundary markers or warnings to the agent to disregard instructions embedded in these external data sources. 3. Capability inventory: The skill allows for the execution of shell commands (e.g.,npm test,npm run build) and modification of repository code and tests. 4. Sanitization: There are no instructions for sanitizing or validating the commands discovered from repository metadata before execution. - [COMMAND_EXECUTION]: The skill encourages the agent to run commands found within the project's environment (e.g.,
npm test,npm run lint). While this is standard for a testing workflow, it relies on the security and integrity of the project's configuration files, which are potentially untrusted if the repository is public or open to external contributions.
Audit Metadata