← Back to list

agent-workflow
by aiskillstore
Security-audited skills for Claude, Codex & Claude Code. One-click install, quality verified.
⭐ 102🍴 3📅 Jan 23, 2026
SKILL.md
name: agent-workflow description: Multi-agent development workflow system. Load when orchestrating development tasks, spawning subagents, or managing workflow phases. allowed-tools: Read, Task, Glob
Agent Workflow System
This skill provides the multi-agent workflow orchestration system.
When to Use
Load when:
- Starting a new development task
- Spawning subagents for parallel work
- Managing workflow phase transitions
- Coordinating multi-agent collaboration
Core Concepts
Workflow Phases
- REQUIREMENTS - PM gathers and clarifies user needs
- ARCHITECTURE - Architects design solutions in parallel
- IMPLEMENTATION - Executor manages parallel implementers
- VERIFICATION - QA validates the implementation
- REFLECTION - Capture learnings for evolution
Agent Hierarchy
Tier-1 (can spawn subagents):
- PM - Requirements gathering
- Architect - Technical design
- Roundtable - Design synthesis
- Executor - Implementation orchestration
- QA - System verification
Tier-2 (focused workers):
- Explorer - Fast codebase scouting
- Implementer - Focused code writing
- Verifier - Independent verification
- Tester - Test execution
- Contract Resolver - Handle blocked tasks
- Reflector - Extract learnings
- Evolver - System improvement
Agent Definitions
Tier-1 Agents:
- @.claude/agents/pm.md
- @.claude/agents/architect.md
- @.claude/agents/roundtable.md
- @.claude/agents/executor.md
- @.claude/agents/qa.md
Tier-2 Agents:
- @.claude/agents/explorer.md
- @.claude/agents/implementer.md
- @.claude/agents/verifier.md
- @.claude/agents/tester.md
- @.claude/agents/contract-resolver.md
- @.claude/agents/reflector.md
- @.claude/agents/evolver.md
Spawning Pattern
Custom Agents (Dotagent System)
Use lowercase names matching agent definitions:
Task(
subagent_type: "explorer", # Fast codebase scouting
model: "haiku",
prompt: "Query: ... Output: ..."
)
Task(
subagent_type: "implementer", # Focused code writing
model: "sonnet",
prompt: "Task: ... Boundaries: ... Output: ..."
)
Task(
subagent_type: "pm", # Requirements gathering
model: "sonnet",
prompt: "Request: ... Output: memory/reports/demand.json"
)
Built-in Agents (Claude Code)
These are different from custom agents above:
Task(
subagent_type: "Explore", # Built-in: quick codebase exploration
model: "haiku",
prompt: "Find all API endpoints..."
)
Task(
subagent_type: "Plan", # Built-in: planning agent
model: "sonnet",
prompt: "Plan the implementation..."
)
Subagent Type Reference
| Custom (Dotagent) | Model | Purpose |
|---|---|---|
pm | sonnet | Requirements gathering |
architect | opus | Technical design |
roundtable | opus | Design synthesis |
executor | sonnet | Implementation orchestration |
qa | sonnet | System verification |
explorer | haiku | Fast codebase scouting |
implementer | sonnet | Focused code writing |
verifier | haiku | Independent verification |
tester | haiku | Test execution |
contract-resolver | sonnet | Handle blocked tasks |
reflector | sonnet | Extract learnings |
evolver | opus | System improvement |
| Built-in (Claude Code) | Purpose |
|---|---|
Explore | Quick codebase exploration |
Plan | Planning and analysis |
general-purpose | Flexible multi-step tasks |
Workflow State
@workflows/main.json @memory/state/phase.json
Score
Total Score
60/100
Based on repository quality metrics
✓SKILL.md
SKILL.mdファイルが含まれている
+20
○LICENSE
ライセンスが設定されている
0/10
○説明文
100文字以上の説明がある
0/10
✓人気
GitHub Stars 100以上
+5
✓最近の活動
1ヶ月以内に更新
+10
○フォーク
10回以上フォークされている
0/5
✓Issue管理
オープンIssueが50未満
+5
✓言語
プログラミング言語が設定されている
+5
✓タグ
1つ以上のタグが設定されている
+5
Reviews
💬
Reviews coming soon
