claude-stream-json

Parse and interpret JSONL stream output from the Claude Code CLI.

62|7|Updated Feb 7, 2026
One-click install
npx skills add https://github.com/littlebearapps/untether --skill claude-stream-json
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: claude-stream-json
Source: https://github.com/littlebearapps/untether/tree/main/.claude/skills/claude-stream-json
Command: npx skills add https://github.com/littlebearapps/untether --skill claude-stream-json

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill provides a structured way to consume and interpret the JSONL output stream from the Claude Code CLI, enabling seamless integration and control of AI coding agents.

Core Features & Use Cases

  • Stream Processing: Parses stream-json output from Claude Code, handling various event types like system initialization, assistant/user messages, and tool usage.
  • Control Channel Integration: Manages interactive permission modes, allowing users to approve or deny tool actions via Telegram.
  • Use Case: When an AI agent needs to execute a bash command, this Skill interprets the tool_use event, presents a permission prompt to the user via Telegram, and processes the tool_result upon completion.

Quick Start

Use the claude-stream-json skill to process the output from a Claude Code CLI session.

Frequently Asked Questions about claude-stream-json

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

FAQPage Schema
How do I parse Claude Code CLI stream-json output for AI agent integration?

To parse Claude Code CLI stream-json output, process the JSONL stream by interpreting event types like system initialization, assistant messages, and tool usage to enable seamless AI agent integration and control.

What is the control channel in Claude Code stream-json protocol?

The control channel in Claude Code stream-json protocol manages interactive permission requests, allowing users to approve or deny tool actions, such as bash commands, and processes tool results upon completion.

How do I handle tool_use events when processing Claude Code CLI streams?

To handle tool_use events in Claude Code CLI streams, intercept the event, present an interactive permission prompt to the user, and then process the corresponding tool_result event upon execution completion.

Can I manage interactive permission requests for Claude Code CLI via Telegram?

Yes, you can manage Claude Code CLI interactive permission requests via Telegram by utilizing control channel communication to present approval prompts and process tool execution results remotely.

What event types are included in Claude Code stream-json output?

Claude Code stream-json output includes event types such as system initialization, assistant and user messages, and tool usage, which require structured parsing to interpret session control flow.