القائمة
MCP-based skill improvement during retrospective - research and update agent skills
يُحدَّد Prompt الذي يراجع المصدر أولًا بشكل افتراضي. يمكنك التبديل إلى أمر مباشر أو تنزيل نسخة محلية.
اقرأ SKILL.md وأي ملفات مرافقة يعرضها SkillsMP قبل أن تقرر التثبيت.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
يتجاوز الأمر المباشر Prompt المخصّص للمراجعة. افحص المصدر قبل تشغيله.
npx skills add https://github.com/feliperyba/ralph-orchestra --skill pm-improvement-skill-researchيبقى الأمر في سطر واحد. مرّر أفقيًا لمراجعته كاملًا قبل النسخ.
تفضّل نسخة محلية؟ نزّل الملفات المتاحة حاليًا لدى SkillsMP.
Complete Developer workflow orchestration - task research sequence, implementation flow, validation gates, PRD synchronization, exit conditions.
Complete Game Designer workflow - skill invocation protocol, GDD creation, playtest flow with GDD review, design sessions. MUST load before starting assignments.
Complete PM Coordinator workflow - task assignment, project orchestration, PRD management, worker coordination. Use proactively when starting PM agent work.
استنادا إلى تصنيف SOC المهني
عرض SKILL.md
| name | pm-improvement-skill-research |
| description | MCP-based skill improvement during retrospective - research and update agent skills |
| category | improvement |
"Continuous learning – improve ALL agent capabilities based on real-world experience."
This is NOT optional. After synthesizing every retrospective, you MUST:
currentTask.status to "skill_research"agents/*/AGENT.md (process improvement)agents/*/skills/*.md (knowledge addition)agents/*/SKILLS.md (core competencies)./.claude/settings.*.json (MCP tool configuration)currentTask = null and assign next taskMinimum per retrospective: At least FIVE skill files updated (one per agent: PM, Developer, Tech Artist, QA, Game Designer).
PM Self-Improvement: The PM MUST also improve at least ONE of its OWN skills each retrospective. See pm-improvement-self-improvement.
Use during skill_research phase when:
## Retrospective: Skill Improvement Check
1. Identify skill gaps for ALL FIVE agents (PM, Developer, Tech Artist, QA, Game Designer)
2. Use MCP (WebSearch, fetch) to research best practices
3. Update at least ONE skill file per agent
4. PM must improve at least ONE of its own skills
5. Document changes in action items
Use this matrix to identify which skills to improve for each agent:
| Agent | Skill Areas to Improve | When to Trigger |
|---|---|---|
| PM | pm-organization-task-selection, pm-retrospective-facilitation, pm-organization-prd-reorganization, pm-improvement-self-improvement, shared-worktree, AGENT.md | Task assignment issues, retrospective gaps, PRD not reorganized, agent workspace conflicts |
| Developer | dev-r3f-r3f-fundamentals, dev-typescript-basics, dev-r3f-r3f-physics, AGENT.md | Code quality issues, failed builds, type errors, worktree violations |
| Tech Artist | ta-r3f-materials, ta-shader-development, ta-assets-workflow, ta-ui-polish, AGENT.md | Visual quality issues, shader errors, asset workflow problems, worktree violations |
| QA | qa-validation-workflow, qa-browser-testing, qa-reporting-bug-reporting, AGENT.md | Missed bugs, incomplete validation, test failures, build conflicts from parallel agents |
| Game Designer | gd-gdd-creation, gd-thermite-integration, gd-validation-playtest, AGENT.md | GDD unclear, playtest issues, design violations |
| Signal | Action |
|---|---|
| Developer asked many clarifications | Improve Developer SKILLS.md with patterns |
| QA missed edge cases | Add testing patterns to QA SKILLS.md |
| Performance issues discovered | Research and add optimization skill |
| New library/API used | Create reference doc in agent's references/ |
| Anti-pattern repeated | Add explicit anti-pattern section |
| PM task assignment struggled | Improve PM task-selection.md |
| PRD not reorganized | Improve PM prd-reorganization.md |
| GDD gaps in implementation | Improve Game Designer gdd-creation.md |
| Build conflicts from parallel agents | Improve PM git-worktree-coordination.md |
| QA reports unexpected changes | Verify agents using separate worktrees |
During retrospective synthesis, identify gaps for ALL FIVE agents:
### Skill Gap Analysis
**PM Gaps**:
- [ ] Missing: PRD reorganization after GDD updates
- [ ] Missing: Git worktree coordination for parallel agents
- [ ] Unclear: Task prioritization algorithms
- [ ] Missing: Risk assessment procedures
**Developer Gaps**:
- [ ] Missing: R3F instancing patterns
- [ ] Missing: Git worktree setup/verification procedures
- [ ] Unclear: Shader uniform management
- [ ] Missing: TypeScript error patterns
**Tech Artist Gaps**:
- [ ] Missing: R3F material optimization patterns
- [ ] Missing: Git worktree setup/verification procedures
- [ ] Unclear: Shader performance profiling
- [ ] Missing: Asset workflow best practices
**QA Gaps**:
- [ ] Missing: WebGL performance profiling
- [ ] Unclear: Mobile device testing
- [ ] Missing: Browser compatibility patterns
**Game Designer Gaps**:
- [ ] Missing: Playtest validation procedures
- [ ] Unclear: GDD-to-PRD task extraction
- [ ] Missing: Thermite editor integration patterns
**Domain Gaps**:
- [ ] Need: Physics collision layers reference
- [ ] Need: Material comparison table
Use available MCP tools to research:
## Research Tasks
### Reference URLs to Fetch
- https://agent-skills.md/skills/Bbeierle12/Skill-MCP-Claude/r3f-fundamentals
- https://agent-skills.md/skills/wollfoo/setup-factory/threejs
- https://agent-skills.md/skills/ovachiever/droid-tings/threejs-graphics-optimizer
- https://github.com/bmad-code-org/BMAD-METHOD
### Web Search Queries
- "React Three Fiber best practices 2026"
- "Three.js mobile optimization techniques"
- "Rapier physics collision layers guide"
Create or update skill files based on research:
Files to Update (PICK AT LEAST ONE PER AGENT):
PM Agent (MUST improve at least one):
agents/pm/AGENT.md — Core PM behavior./.claude/skills/pm-organization-task-selection/ — Task assignment./.claude/skills/pm-retrospective-facilitation/ — Retrospective process./.claude/skills/pm-organization-prd-reorganization/ — GDD-to-PRD extraction./.claude/skills/pm-improvement-self-improvement/ — PM self-improvementDeveloper Agent:
agents/developer/AGENT.md — Core developer behavior./.claude/skills/dev-r3f-r3f-fundamentals/ — R3F patterns./.claude/skills/dev-typescript-basics/ — TypeScript best practices./.claude/skills/dev-r3f-r3f-physics/ — Physics integrationQA Agent:
agents/qa/AGENT.md — Core QA behavior./.claude/skills/qa-validation-workflow/ — Testing process./.claude/skills/qa-browser-testing/ — Browser validation./.claude/skills/qa-reporting-bug-reporting/ — Bug documentationGame Designer Agent:
agents/gamedesigner/AGENT.md — Core Game Designer behavior./.claude/skills/gd-gdd-creation/ — GDD writing./.claude/skills/gd-thermite-integration/ — Thermite editor./.claude/skills/gd-validation-playtest/ — Playtest processTech Artist Agent:
agents/techartist/AGENT.md — Core Tech Artist behavior./.claude/skills/ta-r3f-materials/ — R3F materials./.claude/skills/ta-shader-development/ — Shader creation./.claude/skills/ta-assets-workflow/ — Asset pipeline./.claude/skills/ta-ui-polish/ — Visual polishShared Skills:
./.claude/skills/shared-*/ — Orchestration skillsUpdate Template:
---
name: { { skill-name } }
description: { { one-line with triggers } }
category: { { development|validation|optimization } }
---
# {{Skill Title}}
## When to Use
- {{trigger 1}}
- {{trigger 2}}
## Quick Start
{{minimal example}}
## Anti-Patterns
❌ **DON'T:** {{common mistake}}
✅ **DO:** {{best practice}}
## Reference
- {{external-url}} — {{description}}
Adjust skill depth based on PRD complexity:
| PRD Size | Skill Update Depth |
|---|---|
| 1-5 tasks | Minimal — add anti-patterns only |
| 6-15 tasks | Standard — update relevant skills |
| 16+ tasks | Deep — create new skill files, references |
❌ DON'T:
✅ DO:
During skill_research phase, verify: