القائمة
Delegates coding tasks to the Gemini CLI agent. Use when asked to use gemini, gemini-cli, or to spawn external agents for parallel work on multiple issues.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
This skill helps an LLM generate correct AxAgent tuning and evaluation code using @ax-llm/ax. Use when the user asks about agent.optimize(...), judgeOptions, eval datasets, optimization targets, saved optimizedProgram artifacts, or recursive optimization guidance.
This skill helps an LLM generate correct AxAgent code using @ax-llm/ax. Use when the user asks about agent(), child agents, namespaced functions, discovery mode, shared fields, llmQuery(...), RLM code execution, recursionOptions, or agent runtime behavior. For tuning and eval with agent.optimize(...), use ax-agent-optimize.
This skill helps an LLM generate correct AI provider setup and configuration code using @ax-llm/ax. Use when the user asks about ai(), providers, models, presets, embeddings, extended thinking, context caching, or mentions OpenAI/Anthropic/Google/Azure/Groq/DeepSeek/Mistral/Cohere/Together/Ollama/HuggingFace/Reka/OpenRouter with @ax-llm/ax.
This skill helps an LLM generate correct AxFlow workflow code using @ax-llm/ax. Use when the user asks about flow(), AxFlow, workflow orchestration, parallel execution, DAG workflows, conditional routing, map/reduce patterns, or multi-node AI pipelines.
This skill helps an LLM generate correct AxGen code using @ax-llm/ax. Use when the user asks about ax(), AxGen, generators, forward(), streamingForward(), assertions, field processors, step hooks, self-tuning, or structured outputs.
This skill helps an LLM generate correct AxGEPA optimization code using @ax-llm/ax. Use when the user asks about AxGEPA, GEPA, Pareto optimization, multi-objective prompt tuning, reflective prompt evolution, validationExamples, maxMetricCalls, or optimizing a generator, flow, or agent tree.
| name | delegating-to-gemini-cli |
| description | Delegates coding tasks to the Gemini CLI agent. Use when asked to use gemini, gemini-cli, or to spawn external agents for parallel work on multiple issues. |
Orchestrates work delegation to the Gemini CLI agent for parallel task execution.
Verify gemini CLI is installed:
which gemini && gemini --version
Always specify the model explicitly with -m. Only use Gemini 3 models:
gemini -m gemini-3-pro-preview "task description"
Allowed models:
gemini-3-pro-preview - Best for complex coding tasks, refactoring, multi-file changesgemini-3-flash - Fast for simpler tasks, single-file fixes, quick editsNever use: gemini-2.5-pro, gemini-2.0-flash, or any 2.x models.
gemini -m gemini-3-pro-preview -y "Fix the bug in src/foo.rkt on line 42. Read the file first, then implement the fix and run tests."
Use the Task tool to spawn subagents, each invoking gemini:
<Task>
<description>Fix Issue #1: Description</description>
<prompt>Use the gemini CLI to fix Issue #1.
Run this command:
```bash
cd /path/to/project && gemini -m gemini-3-pro-preview -y "Detailed task description with:
1. Files to read
2. Specific changes needed
3. Tests to run after"
Report back what changes were made.
### Key Flags
| Flag | Purpose |
|------|---------|
| `-m MODEL` | Specify model (required for consistent results) |
| `-y` | Auto-accept all actions (YOLO mode) - use for automated tasks |
| `-d` | Debug mode for troubleshooting |
| `-s` | Sandbox mode for safer execution |
| `--approval-mode auto_edit` | Auto-approve edits only, prompt for other actions |
## Task Description Best Practices
Structure prompts for gemini with:
1. **Context**: What file(s) to read first
2. **Requirements**: Numbered list of specific changes
3. **Verification**: Test command to run after changes
4. **Scope**: Be explicit about what NOT to change
Example:
"Fix the ANSI parser in src/tui/text/ansi.rkt to support 256-color codes.
Requirements:
After changes, run: raco test src/tui/text/ansi.rkt"
## Creating GitHub Issues First
Before delegating, create detailed issues with `gh`:
```bash
gh issue create \
--title "[Component] Brief description" \
--label "bug" \
--body 'Detailed markdown with:
- Problem description
- File locations with line numbers
- Root cause analysis
- Proposed solution
- Acceptance criteria'
Then reference issues in gemini prompts for traceability.
gh issue create including file:line referencesIf gemini hits quota limits:
The model is overloaded. Please try again later.
Fallback options:
gemini-3-flash if using pro (or vice versa)Always verify gemini's work:
# Build check
raco make main.rkt
# Run tests
raco test src/
# Check for uncommitted changes
git status
git diff --stat
# 1. Create issues
gh issue create --title "[TUI] Fix scroll bug" --label "bug" --body '...'
# 2. Spawn subagent with gemini task
# (via Task tool in Amp)
# 3. Verify results
raco test src/tui/
git diff