beam-connect_skill

Overview

This skill connects to a Beam AI workspace and acts as a single entry point for agent management, task orchestration, analytics, and debugging. It validates configuration, discovers and caches agents, and routes user requests to the appropriate Beam operation. The meta-skill coordinates pre-flight checks and hands off to specialized workflows for tasks, graphs, analytics, and optimization.

How this skill works

On first use it runs a mandatory configuration check to validate API key and workspace ID, then prompts the user to provide or store credentials if needed. It can list and cache workspace agents, retrieve agent graphs, create and monitor tasks, fetch analytics, and run node tests or tool optimizations. When a user command matches known triggers or a cached agent name, the skill routes the request to the correct workflow and returns human-friendly summaries and actionable next steps.

When to use it

• When you mention “beam”, “beam agent”, or any Beam-specific phrase
• To list or inspect agents in your workspace
• To create, run, or monitor an agent task
• To fetch agent analytics or performance metrics
• To debug failed tasks, retry, or approve HITL results

Best practices

• Always run the pre-flight config check before operations to avoid auth and workspace errors
• Cache agent lists for the session so users can reference agents by name
• Prompt for and persist BEAM_API_KEY and BEAM_WORKSPACE_ID in a secure .env when setup is required
• Offer to monitor a newly created task and provide concise status updates
• Map common user phrases to workflows to minimize back-and-forth and speed execution

Example use cases

• "List my beam agents" → validate config, list and cache agents, present quick actions
• "Create task for Customer Support: Process latest ticket" → create task and offer progress monitoring
• "Get analytics for Email Processor" → return recent metrics and trends for the agent
• "Retry failed task AGE-785" → fetch task details and retry or provide remediation steps
• "Show agent graph for Report Generator" → display nodes, connections, and entry/exit points