Reticle
An MCP server that checks whether the agent is really done: drives the real app and returns pass, fail or couldn't tell with the file and line
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Edits agent configs and the project build config, restarts the dev server
- Drives the live app and reads its state, including network and store
Install
Manual install
npm install -g @reticlehq/server
npx @reticlehq/server setup mcpThen run npx @reticlehq/server init in the project. For Codex the installer prints lines to paste manually.
This is third-party code. Review the repository files before installing.
What it does
Reticle checks whether the agent's claim of being done matches reality. It drives your live app, reads what actually happened and returns a verdict of pass, fail or couldn't tell with the file and line to fix. Reticle installs as an MCP server and registers with many coding agents via an installer, then an init command wires it into a specific project: it installs a dev-only SDK, edits the build config, restarts the dev server and confirms a session with the app really connected. It ships adapters for React, the browser, Electron, Tauri and build tools.
Who it is for. For developers and QA who want the agent to prove the result in the live app rather than by claim.
Good fit when
- You need to verify that the agent's change actually works in the app
- You need a verdict with the file and line to fix
- You want the agent to read the app's real state, not just the screen
Not a fit when
- The app is not a web or desktop app in the supported environments
- You are not ready to install a dev-only SDK and change the project build config
Example request
Use Reticle to check whether the login form really works after the change, and show the file and line if notLimitations
The installer registers the MCP server with every reachable agent and edits their configs, and for Codex it prints lines to paste manually. The init step changes the project build config and restarts the dev server. A running dev server is required. The license is non-standard (Apache-2.0 with FSL), so check the terms.
How to disable. Remove the reticle entry from the agent configs and uninstall the @reticlehq/server package; revert the build config changes and the dev-only SDK in the project.
MCP
- Transport
- stdio
- Authentication
- not required
Security check
- Edits agent configs and the project build config, restarts the dev server
- Drives the live app and reads its state, including network and store
README in short
The README presents Reticle as a tool that checks whether the agent is really done: it drives the live app, reads real state and returns a verdict of pass, fail or couldn't tell with a file and line. The installer puts reticle on PATH and registers the MCP server with every reachable agent, printing paste lines for Codex. The init step wires the server into the project, installs a dev-only SDK, edits the build config and restarts the dev server. It ships adapters for React, the browser, Electron, Tauri and build tools. Install is also via npm and /plugin marketplace add.
FAQ
How is this different from ordinary browser tests?
Reticle drives the live app and reads its real state, including network and store, and returns a verdict with a file and line rather than only a screenshot.
What does the init step do?
It installs a dev-only SDK, edits the build config, restarts the dev server and confirms a session with the app really connected.
Related
A CLI for testing and red teaming LLM applications: compares models, runs automated CI checks and finds vulnerabilities
Playwright CLI
playwright-cli
Microsoft's official Playwright CLI with an agent skill: drive a browser through short commands without heavy MCP schemas
The official MCP server debugger: web UI, automation CLI and terminal UI in one package
SonarSource's official MCP server: quality and security issues, quality gates and code analysis from SonarQube Server and Cloud