agentmods.dev

CodingAgentExplorer CLAUDE.md

A local reverse proxy and dashboard for AI coding agents such as Claude Code. It forwards requests to language-model and MCP services, records their traffic including streamed responses, and shows activity in a web dashboard.

In plain words
What is it for?
Use it to route Claude Code or MCP traffic through local ports, monitor it in the dashboard, build or run the .NET application, and publish platform-specific files.
Why use it?
It provides a central local view of requests and responses passing between the coding agent and its services. This can help developers inspect agent activity while developing or troubleshooting.

Instructions file

49 2d ago A scan 1,131 tokens original MIT
View the source file tndata/CodingAgentExplorer
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/tndata/codingagentexplorer/claude-md
Clone the repo
git clone --depth 1 https://github.com/tndata/CodingAgentExplorer
Per session 1,131 This file is loaded in full into every session.
When invoked 1,131 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01131 $0.01131
Opus 5 $0.00566 $0.00566
Sonnet 5 $0.00226 $0.00226
Haiku 4.5 $0.00113 $0.00113

Measured 2d ago against content hash 612af6530134, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

CodingAgentExplorer CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 94 lines

How it starts

The opening of the file, as written. The whole thing — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CodingAgentExplorer

ASP.NET Core reverse proxy using YARP that sits between AI coding agents (such as Claude Code) and their LLM API endpoints. Captures all API traffic including streaming SSE responses and exposes a real-time web dashboard via SignalR.

Architecture

  • Port 8888 (localhost): YARP reverse proxy forwarding to LLM API (HTTP)
  • Port 9999 (localhost): YARP reverse proxy forwarding to MCP server (HTTP), destination configured at runtime
  • Port 5000 (localhost): Dashboard (HTTP)
  • Port 5001 (localhost): Dashboard (HTTPS, auto-launches browser)
  • Single project, single NuGet dependency (Yarp.ReverseProxy)

Build & Run

dotnet build
dotnet run

Publish

Windows:

publish.bat

macOS / Linux:

bash publish.sh

Each script builds both projects for the current platform. Outputs to Published/ (gitignored):

  • Published\CodingAgentExplorer\ - framework-dependent, requires .NET 10 runtime
  • Published\HookAgent\HookAgent.exe - single-file, current platform (Windows: win-x64; macOS/Linux: detected at build time)

Usage with Claude Code

Set the API base URL to point at the proxy:

export ANTHROPIC_BASE_URL=http://localhost:8888

Then use Claude Code normally. All requests flow through the proxy and appear on the dashboard at https://localhost:5001.

MCP Observer

The MCP Observer proxies traffic between Claude Code and any MCP server on port 9999. The destination URL is configured at runtime via the dashboard at https://localhost:5001/mcp/index.html.

Register the proxy as an MCP server in Claude Code:

claude mcp add --transport http mcp_proxy http://localhost:9999

The destination URL is stored in McpProxyConfig (singleton) and triggers a YARP config reload via IChangeToken when updated. SSE keep-alive GET requests to / with no body are silently dropped and not stored.

Project Structure

  • Program.cs - App setup: YARP, SignalR, dual-port Kestrel, API endpoints
  • Models/ - DTOs: ProxiedRequest, ClaudeRequestBody, SseEvent, HookEvent, McpDestinationRequest
  • Services/RequestStore.cs - In-memory circular buffer (ConcurrentQueue, max 1000)
  • Services/HookEventStore.cs - In-memory store for hook events
  • Services/McpRequestStore.cs - In-memory store for MCP requests (max 500)
  • Services/McpProxyConfig.cs - Holds the runtime MCP destination URL, signals YARP on change
  • Proxy/CaptureTransformProvider.cs - YARP ITransformProvider for request/response capture
  • Proxy/DynamicProxyConfigProvider.cs - Dynamic YARP config for Claude (8888) and MCP (9999) routes
  • Hubs/DashboardHub.cs - SignalR hub for real-time dashboard updates
  • wwwroot/ - Dashboard SPA (vanilla HTML/JS/CSS + SignalR client)
  • wwwroot/mcp/ - MCP Observer page
  • HookAgent/ - Single-file CLI tool used as a Claude Code hook command
  • publish.bat - Publishes both projects to Published/ (Windows, win-x64 HookAgent)
  • publish.sh - Publishes both projects to Published/ (detects current platform for HookAgent)

Read the full file on GitHub · 94 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 94 lines · 1,131 tokens per session scan A 612af6530134

Subscribe to this mod's changes

CodingAgentExplorer CLAUDE.md is an instructions file published in the GitHub repository tndata/CodingAgentExplorer (49 stars, last pushed 2d ago), licensed MIT. It adds 1,131 tokens to every session, about $0.0057 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Analytics and metricsData and AIPlanningagentic-aiclaude-codeclaudecodedotnet
Related

Other instructions, from other repositories

LitosAiCodingAgent CLAUDE.md

Instructions for nitinmms/LitosAiCodingAgent, covering claude.md and avalonia (litos.gui).

nitinmms/LitosAiCodingAgent · 264 tokens

pm CLAUDE.md

Claude Code instructions for piotrjura/pm, a project described as: Operating discipline for AI coding agents - persistent memory, enforced workflow, recorded decisions.

piotrjura/pm · not measured

intelligent-terminal rust.instructions.md

Concise Rust coding conventions for this repository.

microsoft/intelligent-terminal · 435 tokens

Browser4 CLAUDE.md

Instructions for platonai/Browser4, covering browser4 — project context for claude, architecture, key dispatch chain (cli → browser), batch commands and e2e test structure.

platonai/Browser4 · 1,205 tokens

crystal CLAUDE.md

Instructions for stravu/crystal, covering crystal - multi-session claude code manager, project overview, references, implementation status: ✅ complete and ✅ implemented features.

stravu/crystal · 6,391 tokens

Fluence.Wpf AGENTS.md

Instructions for sintaxasn/Fluence.Wpf, covering fluence.wpf - developer handbook, 1. project overview, solution layout, clr namespaces and 2. coding standards.

sintaxasn/Fluence.Wpf · 13,463 tokens