Back to feed
Dev.to
Dev.to
7/23/2026
Parse transcript.jsonl Directly to See What's Actually Being Called

Parse transcript.jsonl Directly to See What's Actually Being Called

Short summary

A practical script that parses Claude Code's transcript.jsonl files to tally how many times each Skill, Agent, and MCP server was invoked, going beyond the /usage command's cost-only reporting. The script uses Python embedded in Bash to scan JSONL records, filter by time window, and extract tool_use blocks for per-item breakdowns. Useful for cost optimization and environment health checks.

  • •Parses transcript.jsonl to tally per-Skill, per-Agent, and per-MCP-server call counts
  • •Goes beyond Claude Code's /usage which only shows aggregate cost and quota
  • •Includes a ready-to-use Bash+Python script with time-window filtering

Generated with AI, which can make mistakes.

#ai-tools#ai-agents#open-source
Read full article at Dev.to

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

The original title is: "Log Every Agent Invocation: Building Usage Analytics with Claude Code's Stop Hook and transcript_path"
Dev.to
Always-On Cost and Plan-Quota Visibility for Claude Code: Building a Statusline
Dev.to
The original title is: "Fully Custom Claude Code Status Line: A JSON-Driven 3-Line Readout of Quota, Cost, and Health"
Dev.to
Study finds Claude Code sends ~33K tokens before reading user prompt, 4.7x more than OpenCode
Dev.to