BeClaude

ccglass

New
345GitHub TrendingGeneralby jianshuo

See what your coding agent (Claude Code, Codex, Kimi) sends to the model — local proxy + web dashboard

First seen 5/23/2026

Summary

ccglass is a local proxy and web dashboard that lets you see exactly what your coding agent (Claude Code, Codex, Kimi) sends to the model.

  • It helps developers debug, audit, and optimize agent interactions by providing full visibility into the prompts and responses.

Install & Usage

1
Create the agents directory
mkdir -p .claude/agents
2
Save the agent file

Add the configuration to .claude/agents/ccglass.md

3
Invoke with @agent-name
@ccglass

Use Cases

Debug unexpected agent behavior by inspecting the exact prompt sent to the model.
Audit token usage and prompt structure to optimize costs and performance.
Compare prompts across different coding agents to standardize interactions.
Capture and replay agent requests for testing or regression analysis.
Monitor real-time agent activity during complex multi-step coding tasks.
Share prompt logs with teammates to collaborate on agent behavior tuning.

Usage Examples

1

/ccglass start proxy to begin capturing all agent-model traffic

2

Show me the last 5 prompts sent by Claude Code in the dashboard

3

Filter the proxy logs to find all requests containing 'refactor'

View source on GitHub
agent

Security Audits

LicenseUnknownSourceWarnRepositoryPass

Frequently Asked Questions

What is ccglass?

ccglass is a local proxy and web dashboard that lets you see exactly what your coding agent (Claude Code, Codex, Kimi) sends to the model. It helps developers debug, audit, and optimize agent interactions by providing full visibility into the prompts and responses.

How to install ccglass?

To install ccglass: create the agents directory (mkdir -p .claude/agents), then add the config to .claude/agents/ccglass.md. Finally, @ccglass in Claude Code.

What is ccglass best for?

ccglass is a agent categorized under General. It is designed for: agent. Created by jianshuo.

What can I use ccglass for?

ccglass is useful for: Debug unexpected agent behavior by inspecting the exact prompt sent to the model.; Audit token usage and prompt structure to optimize costs and performance.; Compare prompts across different coding agents to standardize interactions.; Capture and replay agent requests for testing or regression analysis.; Monitor real-time agent activity during complex multi-step coding tasks.; Share prompt logs with teammates to collaborate on agent behavior tuning..