Beam Connect

User-facing meta-skill for Beam AI workspace integration.

Purpose

Single entry point for all Beam AI operations:

• Discover workspace agents
• Create and manage tasks
• Monitor analytics and performance
• Debug failed executions
• Optimize tool configurations

Follows the master/connect pattern - references

beam-master

Trigger Phrases

Load this skill when user says:

• "beam" / "beam ai"
• "list agents" / "show beam agents"
• "create beam task" / "run agent task"
• "beam analytics" / "agent performance"
• "beam task status"
• Any agent name from cached context

Pre-Flight Check (ALWAYS RUN FIRST)

Before ANY Beam operation, validate configuration:

python 00-system/skills/beam/beam-master/scripts/check_beam_config.py --json

Handle Config Status

ai_action

proceed_with_operation

prompt_for_api_key

prompt_for_workspace_id

run_setup_wizard

If Setup Needed

I need to set up Beam AI integration first.

To get your credentials:
1. Log into Beam AI (app.beam.ai)
2. Go to Settings → API Keys
3. Create a new API key
4. Also get your Workspace ID from Settings → Workspace

Please provide:
1. Your Beam API key:

After user provides key:

# Write to .env
BEAM_API_KEY=xxx
BEAM_WORKSPACE_ID=workspace-id

# Re-run config check to verify
python 00-system/skills/beam/beam-master/scripts/check_beam_config.py --json

Workflows

Workflow 0: Config Check (Auto)

Trigger: Before any operation Script:

check_beam_config.py --json

Workflow 1: List Agents

Trigger: "list agents", "show beam agents", "my agents"

python 00-system/skills/beam/beam-master/scripts/list_agents.py --json

Display Format:

Found 5 agents in your workspace:

1. Customer Support Agent
   ID: abc-123-def
   Type: beam-os
   Created: 2024-01-15

2. Email Processor
   ID: ghi-456-jkl
   ...

Cache agents for future reference:

• Store agent list in context
• User can reference by name: "run task for Customer Support"

Workflow 2: Get Agent Graph

Trigger: "get agent graph", "show agent workflow", "agent config for X"

python 00-system/skills/beam/beam-master/scripts/get_agent_graph.py --agent-id AGENT_ID --json

Display: Show nodes, connections, entry/exit points

Workflow 3: Create Task

Trigger: "create task", "run agent", "execute agent X"

Required: Agent ID, task query Optional: URLs to parse, context files

python 00-system/skills/beam/beam-master/scripts/create_task.py \
  --agent-id AGENT_ID \
  --query "Task description" \
  --json

Follow-up: Offer to monitor task progress

python 00-system/skills/beam/beam-master/scripts/get_task_updates.py --task-id TASK_ID

Workflow 4: Get Analytics

Trigger: "analytics", "agent performance", "how is X performing"

python 00-system/skills/beam/beam-master/scripts/get_analytics.py \
  --agent-id AGENT_ID \
  --json

Display:

Analytics for Customer Support Agent (Last 30 days)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Tasks: 150 total (+15.5%)
├─ Completed: 135 (+12.3%)
└─ Failed: 15 (-5.2%)

Performance:
├─ Avg Eval Score: 87.5 (+4.5%)
├─ Avg Runtime: 45.7s (-8.7%)
└─ Positive Feedback: 120

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Workflow 5: Task Management

Trigger: "task status", "retry task", "approve task"

Get Task Details:

python 00-system/skills/beam/beam-master/scripts/get_task.py --task-id TASK_ID --json

Retry Failed Task:

python 00-system/skills/beam/beam-master/scripts/retry_task.py --task-id TASK_ID

Approve HITL Task:

python 00-system/skills/beam/beam-master/scripts/approve_task.py --task-id TASK_ID

Provide User Input:

python 00-system/skills/beam/beam-master/scripts/provide_user_input.py \
  --task-id TASK_ID \
  --input "User response"

Rate Task Output:

python 00-system/skills/beam/beam-master/scripts/rate_task_output.py \
  --task-id TASK_ID \
  --node-id NODE_ID \
  --rating positive \
  --feedback "Worked well"

Workflow 6: Test & Update Nodes

Trigger: "test node", "update node config"

Test Node:

python 00-system/skills/beam/beam-master/scripts/test_graph_node.py \
  --agent-id AGENT \
  --node-id NODE \
  --graph-id GRAPH \
  --input '{"key": "value"}'

Update Node:

python 00-system/skills/beam/beam-master/scripts/update_graph_node.py \
  --node-id NODE \
  --objective "New objective"

Workflow 7: Tool Optimization

Trigger: "optimize tool", "improve tool performance"

Start Optimization:

python 00-system/skills/beam/beam-master/scripts/optimize_tool.py --tool TOOL_NAME

Check Status:

python 00-system/skills/beam/beam-master/scripts/get_optimization_status.py --thread-id THREAD

Smart Routing

When user mentions:

Error Handling

Common Errors:

On Error:

1. Load:

   beam-master/references/error-handling.md

2. Display user-friendly message
3. Offer fix if possible

Context Caching

After listing agents, cache for session:

# Store in conversation context
agents_cache = [
    {"name": "Customer Support", "id": "abc-123"},
    {"name": "Email Processor", "id": "def-456"}
]

When user says "run Customer Support", match to cached agent ID.

Integration with Other Beam Skills

This skill can hand off to specialized skills:

beam-list-agents

beam-get-agent-graph

beam-get-agent-analytics

beam-create-agent-task

beam-debug-issue-tasks

Example Interactions

User: "list my beam agents"

AI:

Let me check your Beam configuration first...
✅ Configuration valid

Found 3 agents in your workspace:

1. Customer Support Agent
   ID: abc-123-def
   Description: Handles customer inquiries

2. Email Processor
   ID: ghi-456-jkl
   Description: Processes incoming emails

3. Report Generator
   ID: mno-789-pqr
   Description: Creates weekly reports

What would you like to do? Options:
- "get analytics for Customer Support"
- "create task for Email Processor"
- "show graph for Report Generator"

User: "create task for Customer Support: Process the latest support ticket"

AI:

Creating task for Customer Support Agent...

✅ Task Created!
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Task ID: AGE-785
Status: QUEUED
Query: Process the latest support ticket

Would you like me to monitor this task for updates?

Version

Version: 1.0 Created: 2025-12-11 Status: Production Ready