The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes Defold project files (e.g., .project, .collection, .script) using a provided Python utility and executes build commands. This creates a surface where untrusted project data could contain instructions designed to influence the agent.
Ingestion points: The scripts/check_defold_project.py script reads and parses various resource and script files from the target project directory.
Boundary markers: The instructions do not specify delimiters or techniques to isolate untrusted project data from the agent's control flow.
Capability inventory: The skill enables the agent to execute shell commands (java -jar, python) and run compiled game binaries.
Sanitization: Project content is parsed using regular expressions and displayed/analyzed without specific sanitization or escaping before being presented to the agent.
[COMMAND_EXECUTION]: The skill instructions involve executing shell commands for building projects with Java (bob.jar) and running the resulting game binary under a timeout. This behavior is standard for the skill's intended purpose as a development tool.
[EXTERNAL_DOWNLOADS]: The documentation references and provides download paths for the bob.jar build tool and the headless engine binary from official Defold sources on GitHub and the d.defold.com domain. These are well-known technology service providers.

defold-core