← 返回
AI智能 中文

FIS Architecture

Orchestrate multi-agent workflows with JSON tickets and A2A coordination. Use when delegating tasks between CyberMao (Main) and Worker agents (Engineer/Resea...
通过 JSON 票据和 A2A 协调编排多智能体工作流,用于在 CyberMao 主智能体与工作智能体(工程师/研究员等)之间委派任务。
muselinn
AI智能 clawhub v3.3.1 2 版本 99921.9 Key: 无需
★ 3
Stars
📥 1,219
下载
💾 35
安装
2
版本
#badge#fis#latest#multi-agent#qmd#tickets#workflow

概述

FIS Architecture 3.2 Pro

Multi-agent workflow framework for CyberMao (Main) → Worker coordination using JSON tickets and Discord Forum threads.


When to Use This Skill

Use FIS when:

  • CyberMao (Main) needs to delegate complex tasks to specialized Workers
  • A task requires domain expertise (coding, research, writing)
  • You need to track task status across multiple sessions
  • Multi-step workflows require coordination between agents

Agent Roles:

RoleAgent IDExpertise
---------------------------
ArchitectmainCoordination, task routing, user communication
CodingengineerPython, gprMax, algorithms, data analysis
ResearchresearcherTheory, literature, simulation planning
WritingwriterDocumentation, LaTeX, visualization

Discord Bot Permissions (REQUIRED)

Each agent's Discord bot must have these permissions configured in the Discord server. Without them, Thread creation and messaging will fail silently.

Required Bot Permissions:

  • Send Messages — reply in channels and threads
  • Send Messages in Threads — post inside Forum threads
  • Create Public Threads — create new Forum posts programmatically
  • Read Message History — read thread context
  • Embed Links — send rich embeds in reports
  • Attach Files — upload deliverables

How to configure:

  1. Go to Discord Server Settings → Roles
  2. For each bot role (CyberMao, Researcher, Engineer, Writer), enable the permissions above
  3. Ensure each Forum channel grants these permissions to the relevant bot roles

Verify with:

{ "action": "threadCreate", "channelId": "<forum_channel_id>", "name": "Permission Test" }

If the bot lacks permissions, the discord tool will return an error.


Tool Configuration

ToolPurposePath
---------------------
fis_lifecycle_pro.pyTicket lifecycle (create/status/complete/list)scripts/fis_lifecycle_pro.py
fis_coordinator.pyGenerate delegation templates (CyberMao only)scripts/fis_coordinator.py
fis_worker_toolkit.pySpawn sub-agents, generate reports (Workers only)scripts/fis_worker_toolkit.py

Python Environment: Requires Python 3.8+ with standard library only (no external dependencies).


Core Workflow

Step 1: CyberMao Delegates Task

# Generate ticket + Thread template + A2A command
python3 scripts/fis_coordinator.py delegate \
  --agent engineer \
  --task "Implement GPR signal filter" \
  --forum coding

Output:

  • Ticket ID: TASK_YYYYMMDD_XXX_AGENT
  • Thread template content
  • sessions_send command to notify Worker

Step 2: CyberMao Creates Forum Thread

Use the discord tool to create a Thread in the appropriate Forum channel:

{
  "action": "threadCreate",
  "channelId": "<forum_channel_id>",
  "name": "TASK_xxx: Implement GPR signal filter"
}

The response returns the new Thread ID. Then notify the Worker with the Thread ID:

python3 scripts/fis_coordinator.py notify \
  --ticket-id TASK_xxx \
  --thread-id <new_thread_id>

Execute the generated sessions_send command to notify the Worker.

Step 3: Worker Executes Task

# Check ticket
python3 scripts/fis_lifecycle_pro.py list

# Update status
python3 scripts/fis_lifecycle_pro.py status \
  --ticket-id TASK_xxx --status doing

# Optional: Spawn sub-agent for complex sub-tasks
python3 scripts/fis_worker_toolkit.py spawn \
  --parent-ticket TASK_xxx \
  --subtask "Analyze algorithm complexity"

Worker replies in the Forum Thread using the discord tool:

{
  "action": "threadReply",
  "channelId": "<thread_id>",
  "content": "Task received. Starting execution."
}

Step 4: Worker Reports Completion

# Generate completion report
python3 scripts/fis_worker_toolkit.py report \
  --parent-ticket TASK_xxx \
  --summary "Successfully implemented GPR filter" \
  --deliverables filter.py test_results.json

Execute the generated sessions_send command to notify CyberMao.

Step 5: CyberMao Finalizes

# View report
python3 scripts/fis_coordinator.py report --ticket-id TASK_xxx

# Mark complete
python3 scripts/fis_lifecycle_pro.py complete --ticket-id TASK_xxx

Archive the Thread and report to User in #daily-chat.


Architecture

User/Linn
    ↓
CyberMao (Main) - Architect, coordinator
    ↓ sessions_send + discord threadCreate
Worker (Engineer/Researcher/Writer) - Domain experts
    ↓ (optional) sessions_spawn mode="run"
SubAgent (temporary, background) - Complex sub-tasks

Key Principles:

  1. A2A via sessions_send — Main calls Workers, Workers report back
  2. Ticket tracking — All tasks have JSON tickets in fis-hub/
  3. Programmatic Thread creation — CyberMao creates Forum Threads via discord tool's threadCreate action
  4. SubAgent background modesessions_spawn with mode="run", no new Thread

Commands Reference

fis_lifecycle_pro.py

# Create ticket
python3 scripts/fis_lifecycle_pro.py create \
  --agent engineer --task "Description" --channel-type coding

# Update status (todo/doing/done)
python3 scripts/fis_lifecycle_pro.py status \
  --ticket-id TASK_xxx --status doing --note "Progress update"

# Mark complete
python3 scripts/fis_lifecycle_pro.py complete --ticket-id TASK_xxx

# List active tickets
python3 scripts/fis_lifecycle_pro.py list

# Archive old tickets
python3 scripts/fis_lifecycle_pro.py archive

fis_coordinator.py (CyberMao only)

# Delegate and generate templates
python3 scripts/fis_coordinator.py delegate \
  --agent researcher --task "GPR theory analysis" --forum theory

# Notify Worker after Thread is created
python3 scripts/fis_coordinator.py notify \
  --ticket-id TASK_xxx --thread-id <discord_thread_id>

# View detailed report
python3 scripts/fis_coordinator.py report --ticket-id TASK_xxx

fis_worker_toolkit.py (Workers only)

# Spawn sub-agent (background, no Thread)
python3 scripts/fis_worker_toolkit.py spawn \
  --parent-ticket TASK_xxx --subtask "Complex sub-task description"

# Generate completion report
python3 scripts/fis_worker_toolkit.py report \
  --parent-ticket TASK_xxx \
  --summary "Completion summary" \
  --deliverables file1.py file2.json

Channel Mapping

CategoryForum ChannelWorkerTool Flag
-------------------------------------------
RESEARCH🔬-theory-derivation@Researcher--forum theory
RESEARCH📊-gpr-simulation@Researcher--forum simulation
DEVELOPMENT💻-coding@Engineer--forum coding
WRITING📝-drafts@Writer--forum drafts

Error Handling

If ticket creation fails:

  • Check Python version: python3 --version (need 3.8+)
  • Verify fis-hub/ directory exists and is writable
  • Check disk space

If Thread creation fails:

  • Verify the bot has Create Public Threads permission in the target Forum channel
  • Check that channelId points to a Forum channel (not a regular text channel)
  • Confirm the bot is a member of the server with correct roles

If A2A fails:

  • Verify openclaw.json has agentToAgent.enabled: true
  • Confirm Worker agent ID is in allow list
  • Check Worker session is active

If sub-agent spawn fails:

  • Ensure mode="run" is used (not mode="session")
  • Verify task description is clear and specific

Quality Standards

  1. One Task = One Ticket — Never reuse ticket IDs
  2. Status Updates Required — Workers must update status (TODO → DOING → DONE)
  3. Thread Per Task — Each task gets its own Forum Thread (created via threadCreate)
  4. A2A Confirmation — Always confirm receipt via sessions_send
  5. Archive After Complete — Archive Thread after task completion

Configuration

Required in ~/.openclaw/openclaw.json:

{
  "tools": {
    "agentToAgent": {
      "enabled": true,
      "allow": ["main", "researcher", "engineer", "writer"]
    }
  }
}

Testing

Quick A2A Test

# Test connectivity
sessions_send(sessionKey="engineer", message="A2A test")

Thread Creation Test

{ "action": "threadCreate", "channelId": "<forum_channel_id>", "name": "FIS Test Thread" }

Full Workflow Test

# 1. Create task
python3 scripts/fis_coordinator.py delegate \
  --agent researcher --task "Test task" --forum theory

# 2. Create Forum Thread via discord tool threadCreate

# 3. Notify Worker with Thread ID
python3 scripts/fis_coordinator.py notify \
  --ticket-id TASK_xxx --thread-id <thread_id>

# 4. Execute A2A command

# 5. Complete
python3 scripts/fis_lifecycle_pro.py complete --ticket-id TASK_xxx

FIS 3.2 Pro | Multi-Agent Workflow Framework

版本历史

共 2 个版本

  • v3.3.1 当前
    2026-03-29 03:31 安全 安全
  • v3.2.11
    2026-03-07 01:43

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

ai-intelligence

Self-Improving + Proactive Agent

ivangdavila
自我反思+自我批评+自我学习+自组织记忆。智能体评估自身工作、发现错误并持续改进。
★ 1,362 📥 318,963
ai-intelligence

self-improving agent

pskoett
捕获经验教训、错误和纠正,以实现持续改进。使用时机:(1)命令或操作意外失败;(2)用户纠正……
★ 4,061 📥 799,434

Kimi Quota

muselinn
查询 Kimi API 用量、订阅额度和账户信息。当用户询问"Kimi 额度"、"API 用量"、"剩余次数"、"订阅状态"或需要查看 Kimi Code 使用情况时使用。支持加密保存登录态,首次配置后自动查询。
★ 2 📥 592