The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local CLI tools including entire and git to query repository metadata and checkpoint history. These commands are executed with non-interactive flags (--json, --no-pager) and are strictly within the scope of codebase exploration tasks.
[EXTERNAL_DOWNLOADS]: References are made to https://docs.entire.io/llms.txt, which is a vendor-owned documentation site. This resource follows the llms.txt standard to provide context to the agent and does not involve the execution of remote code.
[PROMPT_INJECTION]: The instructions encourage high-fidelity reporting by requiring the agent to label findings as either 'recorded intent' or 'inferred from code'. This internal logic prevents the agent from hallucinating or misrepresenting historical facts.

using-entire