The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from user bug reports and reproduction steps, which are then used to generate and execute code.
Ingestion points: The agent ingests bug descriptions, intended behavior, and reproduction steps from user input (defined in the Workflow and If a Failing Test Is Impractical sections).
Boundary markers: The instructions lack boundary markers or explicit prompts to help the agent distinguish between data and instructions when processing user bug reports.
Capability inventory: The skill requires the agent to perform file writing (creating tests) and command execution (running tests, regression checks, and validation scripts) as outlined in Workflow steps 3, 4, 6, and 7.
Sanitization: There is no evidence of sanitization or validation of the user's input before it is used to formulate executable tests.
[COMMAND_EXECUTION]: The skill's primary workflow involves the execution of code and scripts in the local environment.
Evidence: The Workflow and Final Response sections explicitly direct the agent to "Run the new test", "Rerun the regression test", and perform "executable regression checks" like targeted scripts or manual reproduction commands.

tdd