The Agent Skills Directory

[Command Execution]: The skill provides templates for diagnostic shell commands, including environment variable inspection (env | grep) and macOS security identity checks (security list-keychains). These patterns are used to verify that necessary build and signing configurations are correctly propagated through multi-component systems.
[Indirect Prompt Injection]: The debugging framework instructs the agent to ingest and analyze external data such as error messages, logs, and stack traces. While this is necessary for root cause analysis, it represents a potential surface where malicious instructions could be embedded in external logs to influence agent behavior. This is a standard consideration for agents that process untrusted output from compilers or external tools.

systematic-debugging