The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill utilizes the @robomotion/sdk/testing library to extract and run JavaScript code from 'Function' nodes within Robomotion flow files. Specifically, the extractFunc and runFunc APIs (documented in reference/function-extraction.md) enable the execution of code parsed from potentially untrusted local files at runtime.
[COMMAND_EXECUTION]: The skill's primary workflow involves executing shell commands, specifically bun test. This command executes the logic contained within *.test.ts files in the working directory. If an attacker can influence the contents of these test files, they can achieve arbitrary code execution on the host system.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data in the form of flow definitions (main.ts) and test scripts. This creates a surface where malicious instructions or code embedded in those files could be executed by the agent.
Ingestion points: The skill loads external files via FlowTester.load(), SubflowTester.load(), and extractFunc() (referenced in SKILL.md and reference/flow-tester.md).
Boundary markers: No specific delimiters or safety warnings are implemented to distinguish between flow data and executable instructions during the loading process.
Capability inventory: The skill possesses the ability to run shell commands (bun test) and execute extracted JavaScript logic (runFunc).
Sanitization: There is no evidence of sanitization or validation performed on the code extracted from Function nodes before it is executed.

testing-flow