gemini-3-prompting

Gemini 3 Prompting

Use this skill when gemini:gemini-rescue needs to ask Gemini for help.

Prompt Gemini like an operator, not a collaborator. Keep prompts compact and block-structured with XML tags. State the task, the output contract, the follow-through defaults, and the small set of extra constraints that matter.

Core rules:

• Prefer one clear task per Gemini run. Split unrelated asks into separate runs.
• Tell Gemini what done looks like. Do not assume it will infer the desired end state.
• Add explicit grounding and verification rules for any task where unsupported guesses would hurt quality.
• Prefer better prompt contracts over raising reasoning or adding long natural-language explanations.
• Use XML tags consistently so the prompt has stable internal structure. Gemini 3 responds well to explicit <thinking> tags and to JSON schemas given inline — use both when the output shape matters.

Default prompt recipe:

• <task>: the concrete job and the relevant repository or failure context.
• <structured_output_contract> or <compact_output_contract>: exact shape, ordering, and brevity requirements.
• <default_follow_through_policy>: what Gemini should do by default instead of asking routine questions.
• <verification_loop> or <completeness_contract>: required for debugging, implementation, or risky fixes.
• <grounding_rules> or <citation_rules>: required for review, research, or anything that could drift into unsupported claims.

When to add blocks:

• Coding or debugging: add completeness_contract, verification_loop, and missing_context_gating.
• Review or adversarial review: add grounding_rules, structured_output_contract, and dig_deeper_nudge.
• Research or recommendation tasks: add research_mode and citation_rules.
• Write-capable tasks: add action_safety so Gemini stays narrow and avoids unrelated refactors.

How to choose prompt shape:

• Gemini has no built-in review mode, so all review jobs go through review or adversarial-review subcommands that already carry the review contract.
• Use task when the task is diagnosis, planning, research, or implementation and you need to control the prompt more directly.
• Use task --resume-last for follow-up instructions on the same Gemini thread. Send only the delta instruction instead of restating the whole prompt unless the direction changed materially.

Use Gemini's built-in subagents when the task benefits from dedicated exploration:

• codebase_investigator (ships with the codebase-investigation skill) — invoke it inside a prompt when Gemini needs to map call sites, trace invariants, or summarize a subsystem before answering. It is configured to use gemini-3.1-pro-preview. Mention it by name in the prompt body (e.g., "use the codebase_investigator subagent to map X before finalizing the fix") when you want Gemini to reach for it instead of guessing from a shallow read.

Working rules:

• Prefer explicit prompt contracts over vague nudges.
• Use stable XML tag names that match the block names from the reference file.
• Do not raise reasoning or complexity first. Tighten the prompt and verification rules before escalating.
• Ask Gemini for brief, outcome-based progress updates only when the task is long-running or tool-heavy.
• Keep claims anchored to observed evidence. If something is a hypothesis, say so.

Prompt assembly checklist:

1. Define the exact task and scope in <task>.
2. Choose the smallest output contract that still makes the answer easy to use.
3. Decide whether Gemini should keep going by default or stop for missing high-risk details.
4. Add verification, grounding, and safety tags only where the task needs them.
5. Remove redundant instructions before sending the prompt.

Reusable blocks live in references/gemini-prompt-blocks.md. Concrete end-to-end templates live in references/gemini-prompt-recipes.md. Common failure modes to avoid live in references/gemini-prompt-antipatterns.md.