cursor
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell scripts to invoke the
cursor-agent(oragent) command-line tool. This is the primary function of the skill to execute coding tasks delegated by the user. - [DYNAMIC_EXECUTION]: The main runner script dynamically generates helper scripts (
monitor.shandcontinue.sh) in the run's artifact directory. These scripts are used to track the progress of the background agent process and allow for task resumption. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided prompts or prompt files and passes them directly to the underlying agent. While this represents a potential attack surface for indirect prompt injection, it is a necessary component of the skill's intended operation and includes no unsafe interpolation or evaluation within the wrapper itself.
- [SAFE]: The skill manages the
CURSOR_API_KEYfor authentication. It implements logic to retrieve the key from environment variables or specific files while explicitly providing mechanisms to prevent the key from being logged or printed to the console.
Audit Metadata