The Agent Skills Directory

[COMMAND_EXECUTION]: The skill reads shell commands from a local configuration file (agents.md) and executes them to verify the generated tests. This execution path is determined by the content of the repository files.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from issue-analysis.md to guide test generation and logic.
Ingestion points: The agent reads issue-analysis.md and agents.md from the repository root or a specified path.
Boundary markers: No specific delimiters or instructions are used to prevent the agent from obeying instructions embedded within the ingested bug reports.
Capability inventory: The skill has the capability to write new files to the repository and execute shell commands for test verification.
Sanitization: No sanitization or validation of the input data from the markdown files is performed before processing.

create-automation