The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to review and interact with existing Flutter test code, which presents a surface for indirect prompt injection where instructions embedded in project files could influence agent behavior.
Ingestion points: The skill reviews unit, widget, and integration tests as part of its core functionality in SKILL.md.
Boundary markers: The instructions lack specific delimiters or warnings to ignore instructions embedded within the code files being analyzed.
Capability inventory: The skill can execute shell commands via the Flutter test runner and perform file system operations (create/delete) as documented in the "Disposable test probes" section.
Sanitization: There are no explicit steps to sanitize or validate the content of reviewed test files before they are processed or executed.
[DYNAMIC_EXECUTION]: The skill generates and executes Dart source files at runtime for diagnostic purposes.
Evidence: The "Disposable test probes" section in SKILL.md and the evals/cases.json file explicitly instruct the agent to create temporary *_test.dart files and run them.
[COMMAND_EXECUTION]: The skill relies on executing host-side commands to perform testing and cleanup tasks.
Evidence: The verification section in SKILL.md instructs the use of the repository's established test command or MCP capabilities. The lifecycle management for disposable probes involves host-side file creation and deletion commands.

flutter-testing