Regressguard MCP server

RegressGuard

Before you commit, know what broke.

When an AI coding agent edits your app it can silently break an API contract — a removed field, a changed status code, a test that now fails — and still report success. RegressGuard records a known-good baseline and tells you (or the agent) exactly what regressed.

It is built to live inside the agent's own loop. RegressGuard ships as an MCP server, so agents like Claude Code and Cursor can verify their own work and self-correct before a human ever sees the diff — zero extra steps. The same engine also runs as a plain CLI for humans and CI.

# Agent-native (primary): the agent calls these as MCP tools in its loop
snapshot → check → status        # see "Agent-native verification (MCP)" below

# Human / CI (also works): two commands, no test-writing, under 15 seconds
rg snapshot   # record the known-good state
rg check      # compare after edits — see what broke

Break → detect → fix → green. Reproduce it yourself: ./demo/demo.sh.

Install

macOS / Linux (recommended)

curl -fsSL https://raw.githubusercontent.com/Bharath-code/regressguard/main/install.sh | sh

Homebrew

brew install Bharath-code/tap/rg

Verify

rg version   # first line must say "RegressGuard"

Have ripgrep installed? ripgrep also ships as rg, and whichever comes first on

PATH wins — rg check could silently run ripgrep instead of RegressGuard. If

rg version doesn't say "RegressGuard", invoke the full path (e.g.

/usr/local/bin/rg). The pre-commit hook and GitHub Action already use absolute

paths and are unaffected; rg doctor flags the collision.

Quickstart (3 minutes)

1. Initialize your project

cd your-project
rg init

RegressGuard detects your test command, framework, and dev server URL automatically.

2. Record the baseline before your AI session

Make sure your dev server is running, then:

rg snapshot

Output:

Snapshot

OK Tests       42 passed, 0 failed       6.8s
OK Routes      6 captured, 2 skipped
OK Schemas     6 hashed

Saved:
  .regressguard/snapshot.json

Next:
  Ask your AI agent to make the code change, then run:
  rg check

3. Run your AI agent

Let Claude Code, Cursor, or Codex make its changes.

4. Check for regressions before committing

rg check

Clean — safe to commit:

Check

OK No regressions detected

  Tests       42 passed, 0 failed
  Routes      6 unchanged
  Timing      within tolerance

Safe to commit.

Regression found — commit blocked:

Check

X 2 regressions detected

  Route                                 Before    After     Change
  GET /api/users                        schema    schema    schema
    - role (string, removed)
    + age (number, added)
  POST /api/user/update                 200       500       status

Likely cause:
  Auth/session behavior or routing changed during the last code edit.

Changed files since snapshot:
  app/api/users/route.ts
  internal/auth/session.go

Next:
  rg check --verbose
  git diff

Commit blocked.

Exit code 1 on critical — works with git hooks and CI.

Git Hook (auto-protect every commit)

rg hook install

Now rg check runs automatically before every git commit. When a critical regression is detected, the commit is blocked with a compact output:

RegressGuard pre-commit

Shortened. The full README is on GitHub.