Choose your path
How to go from "should we try Molar?" to a gated PR and a replayable trace — without getting lost between marketing pages, docs, and the dashboard.
Molar has four products (Cartographer, Guard, Clones, Trace) and one operator UI at app.molar.it. The CLI and MCP server are included with every account.
Three phases
Evaluate Install & connect Operate
───────── ───────────────── ───────
molar.it npx molar-agent init app.molar.it
product landings .molar.md + GitHub dashboard + traces
docs.molar.it first scenario run Guard + Mender loop
Phase 1 — Evaluate (no account required)
Goal: Understand which product fits your team and what Molar actually ships today.
| Step | Where | What to do |
|---|---|---|
| 1 | molar.it | Read the four-product story |
| 2 | Product landing | Deep-dive the product you care about — cartographer.molar.it, guard.molar.it, clones.molar.it, trace.molar.it |
| 3 | Getting started | Pick human vs agent path; confirm fit (clones, GitHub App, BYOK) |
Pick your entry product:
| If your pain is… | Start at… | You'll likely also need… |
|---|---|---|
| "We don't have E2E coverage" | Cartographer | Clones (vendors), Guard (CI gate) |
| "Stripe/Twilio test mode is killing CI" | Clones | Cartographer or your existing tests |
| "Regressions merge anyway" | Guard | Cartographer (scenarios), Trace (debug) |
| "Flaky tests / can't reproduce failures" | Trace | Guard (captures traces on failure) |
Phase 2 — Install & connect
Goal: A project wired to Molar with at least one scenario running against a safe environment.
| Step | Where | What to do |
|---|---|---|
| 1 | app.molar.it | Create account → onboarding (project + optional app URL) |
| 2 | Terminal | npx molar-agent init — scaffolds .molar/ and sample scenarios |
| 3 | Dashboard | Confirm app URL on your Cartographer project |
| 4 | Dashboard or MCP | Discover flows (Scouts or molar_generate_scenarios_from_url) → approve as scenarios |
| 5 | .molar.md | Wire Clones base URLs if your app hits Stripe, email, etc. |
| 6 | CLI or dashboard | molar run (local runner required) or Run from Scenarios |
| 7 | Dashboard | Open the run's Trace — confirm forensic replay works |
Docs to keep open:
- .molar.md config — project config reference
- BYOK — if you bring your own LLM keys
- Clones — vendor clone setup
- Error codes — when something fails
Phase 3 — Operate (dashboard + CI loop)
Goal: Every PR is gated; failures are replayed and optionally auto-fixed.
| Step | Where | What to do |
|---|---|---|
| 1 | Settings → Organization | Install Molar GitHub App on your repo |
| 2 | GitHub branch protection | Require molar/guard status check |
| 3 | app.molar.it/guard/pr-gate | Configure PR gating |
| 4 | app.molar.it/guard/monitoring | Production synthetics (when entitled) |
| 5 | On failure | Open Trace from the run → diff vs last green |
| 6 | Settings → Notifications | Email on gate failures |
Human vs agent onboarding
| Humans | Agents | |
|---|---|---|
| Account | app.molar.it signup | molar agent signup + 6-digit verify |
| Primary UI | Dashboard | MCP + CLI |
| Credentials | PAT from Settings → API Tokens | Agent API key or human-minted PAT |
| First run | Dashboard Run or molar run | MCP molar_run_scenarios or molar run |
See For humans and For agents for full detail.
Related
- For humans — app.molar.it signup, dashboard, GitHub App
- For agents — MCP, self-serve signup, llms.txt
- All-in-one dashboard — app.molar.it route reference