beam-debug-issue-tasks

Diagnose failed Beam.ai agent tasks by analyzing Langfuse traces and generating debug reports.

Updated Jan 22, 2026
One-click install
npx skills add https://github.com/FahadImdad/Nexus-Fahad --skill beam-debug-issue-tasks-fahadimdad
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: beam-debug-issue-tasks
Source: https://github.com/FahadImdad/Nexus-Fahad/tree/main/00-system/skills/beam/beam-debug-issue-tasks
Command: npx skills add https://github.com/FahadImdad/Nexus-Fahad --skill beam-debug-issue-tasks-fahadimdad

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, python-dotenv, and includes scripts (resource) components.

What problem does it solve? When Beam.ai agent tasks fail, stop, or require input, finding the root cause means manually digging through task logs and Langfuse traces. This Skill automates that investigation by fetching issue tasks, analyzing trace reasoning, and producing structured debug reports. ## Core Features & Use Cases - Issue Task Detection: Lists tasks with statuses like FAILED, ERROR, STOPPED, TIMEOUT, or USER_INPUT_REQUIRED across BID and Prod workspaces over configurable lookback periods. - Langfuse Trace Analysis: Extracts reasoning from key spans (ParameterSelection, ExecuteGPT_Tool, NodeSelection:EdgeEvaluation, TaskSuccessCriteriaCheck) to identify root causes, with direct session and trace URLs. - Smart Brevity Debug Reports: Saves markdown reports with headline, takeaway, details, root cause, and fix suggestions to per-agent debug folders. - Use Case: An agent's tasks keep stopping with 'condition failed'. Run the script with the agent ID to see that a checklist evaluation failed because a subfolder parameter was null, then fix the input data. ## Quick Start Ask the AI to debug issue tasks for a specific Beam agent ID, optionally specifying a task ID, workspace, or lookback period in days.

Frequently Asked Questions about beam-debug-issue-tasks

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug failed Beam.ai agent tasks?

Run the debug_issue_tasks.py script with an agent ID to list issue tasks from the last day, or pass --task-id to debug a specific task. The script fetches Langfuse traces and extracts root-cause reasoning from key generation spans.

How do I find why a Beam task stopped or failed using Langfuse?

The script analyzes Langfuse trace observations, focusing on spans like ParameterSelection, ExecuteGPT_Tool, and NodeSelection:EdgeEvaluation. It extracts the last reasoning output and errors, then includes direct session and trace URLs for deeper inspection.

What task statuses are treated as issues in Beam.ai?

Tasks with status FAILED, ERROR, STOPPED, CANCELLED, TIMEOUT, or USER_INPUT_REQUIRED are flagged as issues. Statuses like COMPLETED, IN_PROGRESS, QUEUED, and RUNNING are excluded as normal operation.

Can I debug tasks from both Beam BID and Prod workspaces?

Yes, use the --workspace flag with 'bid' or 'prod' to select the Beam API endpoint and corresponding Langfuse project. Each workspace requires its own API key and workspace ID in the .env file.

Why does the script report no traces found for a task?

This happens when the agent lacks Langfuse integration or the task ID does not match any session in the selected Langfuse project. Verify the agent has tracing enabled and that you selected the correct workspace and project.