القائمة
Launch N parallel research agents that explore different improvement approaches, cross-pollinate ideas, and converge on a unified plan
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
Create a git branch from a base and set its upstream at once, so a later push can never land on the base branch. Use when starting work that needs its own branch.
Create a well-structured GitHub issue with duplicate detection, code verification, and auto-labeling
Root-cause a bug or a failed fix with evidence, not guesses - brainstorm causes, research them in parallel, confirm one with logs before changing code. Use when a reported behavior needs root-causing or a fix attempt has failed.
Implement a GitHub issue interactively - asks about branch, PR target, and phases/steps before starting work
Rename or refactor a symbol safely by searching every naming-case variant across code, tests, docs, configs, and JSON. Use when renaming any identifier so no reference is missed.
Review PR comments, apply or reject each one with evidence, and post replies directly to GitHub
| name | research-agents |
| description | Launch N parallel research agents that explore different improvement approaches, cross-pollinate ideas, and converge on a unified plan |
Launch parallel research agents that each explore a different angle of improvement for a feature or area, then iterate through rounds of cross-pollination until they converge on a unified recommendation.
TARGET: The feature or area to improve (e.g., "menu generation algorithm", "recipe editor UX", "persistence format")AGENTS: Number of agents (default: 4, range: 2-6)ROUNDS: Number of rounds (default: 2)ANGLES: Comma-separated research angles (optional)Understand what's being improved:
git log --oneline -20)Produce a concise Target Summary.
Propose angles tailored to this specific target. Present to the user with AskUserQuestion for confirmation.
Different systems call for different angles:
| System Type | Example Angles |
|---|---|
| Algorithm (menu gen) | Optimization quality, Performance, Configurability, Edge case handling |
| UI/UX (recipe editor) | Usability, Accessibility, State management, Animation/polish |
| Data model (Freezed) | Schema design, Serialization robustness, Migration strategy, Query patterns |
| Architecture | Testability, Separation of concerns, Scalability, Developer ergonomics |
Wait for user confirmation before proceeding.
<target-dir>/planning-workspace/
+-- config.md
+-- target-summary.md
+-- agents/
| +-- agent-1-<name>/brief.md
| +-- agent-2-<name>/brief.md
+-- synthesis/
Launch ALL agents in parallel in a SINGLE message using the research-agent agent definition.
Launch ALL agents again in parallel. Each reads all others' prior round outputs before writing.
Stop early if agents' Top 3 priorities are aligned and disagreements are resolved.
Launch ONE synthesis agent using research-synthesizer.
Present the Executive Summary, highlight tradeoffs requiring user input, and offer to dive deeper or begin implementation.