From gemini
Gemini model prompting guide for effective task delegation from Claude Code
How this skill is triggered — by the user, by Claude, or both
Slash command
/gemini:gemini-promptingThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill only to shape the prompt text before forwarding it to the `gemini-companion` runtime. Do not use it to do independent work, inspect the repository, or draft solutions.
Use this skill only to shape the prompt text before forwarding it to the gemini-companion runtime. Do not use it to do independent work, inspect the repository, or draft solutions.
--thinking-budget for complex reasoning tasks.Use XML tags to structure prompts for clarity:
<role>
Define the persona and operating stance.
</role>
<task>
State what needs to be done. Be specific about inputs and expected outputs.
</task>
<context>
Provide repository context, file contents, error messages, or other relevant information.
</context>
<constraints>
List hard boundaries: what NOT to do, scope limits, safety rules.
</constraints>
<output_contract>
Define the exact output shape. Use JSON schema descriptions or examples.
</output_contract>
auth.go:142" beats "Fix the bug".npx claudepluginhub sakibsadmanshajib/gemini-plugin-cc --plugin geminiCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.