The Agent Skills Directory

[SAFE]: The skill operates as a command dispatcher and does not directly execute shell commands, perform network operations, or access sensitive files.
[SAFE]: Includes explicit instructions to guard against indirect prompt injection by warning the agent that external data such as pull request descriptions or issue content must be treated as untrusted input and not be obeyed as instructions.
[SAFE]: Implements a requirement for explicit user authorization before delegating any actions that involve modifying source code or test files.

test-dispatch