Agent Roast
Offline review that flags agent changes which weaken tests, assertions or CI
- Pricing
- Open source
- Platforms
- CLI, Linux, macOS
- Runs
- Locally
- Version
- 0.28.0
- Listing updated
- 3 days ago
About Agent Roast
Agent Roast is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
agent-roast compares the before and after of a code change and reports edits that make a build pass by weakening the checks rather than fixing the code. Examples include loosened assertions, skipped or deleted tests, lowered coverage thresholds and CI steps allowed to fail.
It is deterministic and runs offline with no model or network. It can be used as a hook for coding agents, a Git hook or a pull request check, and any MCP-capable agent can call it as a read-only tool. When a check was weakened, the agent is sent back once to fix the code.
It is aimed at developers who let coding agents change repositories and want a guard against test tampering.
Install
Run in a terminal
claude mcp add agent-roast -- npx -y agent-roastAdd to claude_desktop_config.json
{
"mcpServers": {
"agent-roast": {
"command": "npx",
"args": [
"-y",
"agent-roast"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"agent-roast": {
"command": "npx",
"args": [
"-y",
"agent-roast"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"agent-roast": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"agent-roast"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"agent-roast": {
"command": "npx",
"args": [
"-y",
"agent-roast"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"agent-roast": {
"command": "npx",
"args": [
"-y",
"agent-roast"
]
}
}
}What it does
- Before and after diffing
- Compares changed test and config files to spot weakened checks.
- Agent hooks
- Hook installation for Claude Code, Codex, Cursor, Gemini CLI, Copilot and Factory Droid.
- PR and CI check
- Runs as a pull request check or in any CI with SARIF output.
- Rule set
- Rules for weakened assertions, skipped tests, lowered guards and trivial asserts.
- Offline operation
- No model, tokens or network are involved.
- Replay mode
- Shows what it would have reported about recent commits.
Good for
- Stop coding agents from skipping failing tests
- Review uncommitted changes for weakened assertions
- Add a pull request check against lowered coverage thresholds
- Audit recent commits for test tampering
- Expose a read-only review tool to MCP agents
Pricing
Open source
Free, MIT license
Agent Roast FAQ
Does it use an AI model?
No. It is deterministic and offline.
How do I try it?
Run npx agent-roast demo for a scripted example, or npx agent-roast to review uncommitted changes.
What languages are covered?
JavaScript/TypeScript, Python and Go get full before/after comparison; skips and placeholder assertions also cover several other languages.
What license is it under?
MIT.
Is Agent Roast free?
Yes. Agent Roast is open source, so you can use and self-host it for free.
How do I add Agent Roast to Claude Code?
Run this command in your terminal: claude mcp add agent-roast -- npx -y agent-roast
What are the best alternatives to Agent Roast?
The closest alternatives to Agent Roast are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to Agent Roast

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Ouroboros
MCP serverSpec-first layer that pins requirements and verifies results of AI coding agent work
Runs locally- Open source
- 6.4K GitHub stars
- Developer tools