OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Evaluation & Observability

Claude Tap

Claude Tap is a local proxy and trace viewer for inspecting API traffic and agent context from coding-agent clients. It displays prompts, conversation history, tool schemas and calls, streaming responses, token usage and request diffs.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Claude Tap — OpenViST project profile in Agent Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Claude Tap during development when you need to see the API context and traffic behind a coding-agent run, or compare requests to investigate a change in behaviour.

Who it is for

  • Developers debugging coding-agent behaviour at the request and context level.
  • Teams or individuals reviewing agent traces locally or exporting a session for review and archiving.

Documented uses

  • Capture API traffic while running a supported coding-agent client.
  • Inspect prompts, conversation history, tool definitions, calls, responses and token usage.
  • Compare adjacent requests to identify changes in prompts, messages, tools or parameters.
  • Export a trace session as a self-contained HTML viewer.
Before you choose

Requirements & limits

The README requires Python 3.11 or later and the coding-agent client being traced. The project is MIT-licensed.

  • Claude Tap can launch a CLI through its proxy or listen to local app transcripts; capture setup varies by client.
  • The viewer includes request diffs, token usage and reconstructed streaming responses, beyond a basic transcript view.
  • The README says common authentication headers are redacted before recording and no hosted dashboard is required.
  • Trace sessions can be exported as self-contained HTML for review or archiving.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Inspect a coding-agent run with claude-tap

Launch a supported coding-agent client through claude-tap and review its API traffic in the local trace viewer.

Before you start

  • Python 3.11 or later.
  • The coding-agent client you want to trace.
  • claude-tap installed; the README recommends uv tool install claude-tap.
  • A client listed as supported by the README, such as Claude Code or Codex CLI.

Steps

  1. Install claude-tap with the README's recommended command: the command below.

    Documented command
    uv tool install claude-tap

  2. Run the desired client through claude-tap; for example, use claude-tap for Claude Code or claude-tap --tap-client codex for Codex CLI.

  3. Use the client to perform the run you want to inspect.

  4. Open the live browser viewer and inspect the captured requests, agent context, responses, token usage, or request differences.

Expected outcome

A local trace session is available for inspecting the run's API traffic and agent context in the viewer.

Guide limitations

  • Requires Python 3.11 or later and the client being traced.
  • Proxy-only setups require starting the client separately and pointing its base URL or proxy settings at the local proxy; consult the client support matrix for exact wiring.
  • The README states that common auth headers are redacted before recording; it does not establish that all sensitive data is removed.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
10 October 2026
Original website title
liaohch3/claude-tap · Official homepage title · retained in its source language
How we review sourcesSuggest a correction