The Agent Skills Directory

[SAFE]: The skill implements a robust, iterative process for improving code quality by adding tests until a configured coverage target is reached. It uses project-native commands and artifacts.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute standard coverage reporting tools such as pytest, jest, and go test. These commands are resolved from the project's own configuration or standard environment defaults, representing normal development activity.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of coverage reports (e.g., XML, JSON, and terminal output). While this presents a theoretical attack surface, the skill limits the risk by specifically parsing numeric metrics and file paths to inform its next action. There are no indications of malicious instruction ingestion.
[DYNAMIC_EXECUTION]: The skill dynamically generates test code based on the source code it analyzes. This is the primary function of the skill and is constrained by a 'default-FAIL' contract and mandatory green-suite verification steps.

coverage-loop