One command. 10 lenders. Sorted best to worst. Stop opening a dozen bank websites — this tool checks them all in 30 seconds with stealth browser automation.

Multi-Lender Mortgage Rate Lookup

Compare mortgage rates across 10 lenders and 2 national benchmarks in a single command. All major banks, credit unions, and online lenders — scraped in parallel, ranked lowest to highest, with day-over-day market tracking.

---

The Problem

Shopping for the best mortgage rate means opening a dozen bank websites, entering your info on each one, waiting for JavaScript to load, and manually comparing numbers. Every. Single. Day.

The Solution

One command. 10 lenders. 2 national benchmarks. Sorted best to worst. Typically completes in 30-90 seconds depending on network speed and lender response times.

The script hits every lender simultaneously using stealth browser automation, extracts the current rate and APR, compares them against Freddie Mac and Mortgage News Daily national averages, tracks how rates moved since yesterday, and gives you a clean ranked report you can pipe to Discord, Slack, email, or any automation pipeline.

---

How We Beat Anti-Bot Protection

Major banks don't want you scraping their rate pages. They use JavaScript-rendered SPAs, Cloudflare protection, bot detection fingerprinting, and rate-limiting to block automated access. A simple requests.get() or even a standard headless browser will get blocked, served empty pages, or fed placeholder text like X.XXX%.

How this tool gets through:

Challenge	How We Solve It
JavaScript-rendered rates	patchright — a stealth-patched Chromium that renders full JS like a real browser
Bot detection fingerprinting	patchright spoofs browser fingerprints (WebGL, canvas, navigator properties) to look like a real user
Rate-limiting / IP blocking	Batches of 4 lenders at a time with retry logic — never hammers a single site
ZIP code gates (Chase)	Auto-detects ZIP input fields, fills them, and clicks "See Rates" before extracting
Session-dependent rates (BofA)	Uses specific promo URLs that serve rates on first load without multi-step navigation
Cloudflare challenges	patchright bypasses standard Cloudflare browser checks without needing API keys

Every lender required a different approach to crack. The URLs, wait times, and extraction patterns in this script are the result of testing each bank individually until we found what works.

If a lender stops working: Bank websites change their structure regularly. If a lender starts returning empty results, use --headed mode to open a visible browser and see what changed. Then update the URL or regex pattern accordingly.

---

What Gets Tracked

10 Lenders

#	Lender	Type
1	Bank of America	Big 4 Bank
2	Wells Fargo	Big 4 Bank
3	Chase	Big 4 Bank
4	Citi	Big 4 Bank
5	Navy Federal Credit Union	Credit Union
6	SoFi	Online Lender
7	US Bank	National Bank
8	Guaranteed Rate	Online Lender
9	Truist	National Bank
10	Mr. Cooper	Largest Servicer

All 10 lenders are fully automated — no manual browser intervention needed.

2 National Benchmarks

Source	Description	Frequency
Freddie Mac PMMS	Official national average from the Primary Mortgage Market Survey — the industry gold standard	Weekly
Mortgage News Daily	Real-time daily rate index tracked by mortgage professionals nationwide	Daily

Every lender's rate is displayed alongside these benchmarks so you can immediately see who's above or below the national average.

Products Compared

• 30-Year Fixed (rate + APR)
• 15-Year Fixed (rate + APR)
• FHA 30-Year (where available)
• VA 30-Year (where available)
• ARM (where available)

---

Sample Output

MORTGAGE RATE COMPARISON — Mar 25, 2026
   10 lenders tracked + 2 benchmarks | 10 reporting now

--- 30-YEAR FIXED -----------------------------------------------

  [BEST] Navy Federal CU         5.625%                     BEST
     Wells Fargo             5.750%  (5.968% APR)
     SoFi                    5.990%
     Freddie Mac (natl avg)  6.220%  ·················· benchmark
     Guaranteed Rate         6.250%  (6.547% APR)
     Truist                  6.350%  (6.510% APR)
     Chase                   6.375%  (6.481% APR)
     US Bank                 6.375%  (6.504% APR)
     Bank of America         6.500%  (6.738% APR)
     MND Index               6.480%  ·················· benchmark
     Citi                    6.625%  (6.750% APR)
     Mr. Cooper              6.750%  (6.820% APR)

  Avg: 6.330%  |  vs yesterday: down 0.010%

--- 15-YEAR FIXED -----------------------------------------------

  [BEST] Navy Federal CU         5.250%                     BEST
     SoFi                    5.375%
     US Bank                 5.490%  (5.760% APR)
     Guaranteed Rate         5.500%  (5.919% APR)
     Freddie Mac (natl avg)  5.540%  ·················· benchmark
     Wells Fargo             5.625%  (5.874% APR)
     Truist                  5.700%  (5.980% APR)
     Bank of America         5.750%  (6.134% APR)
     Chase                   5.875%  (6.012% APR)
     Citi                    5.875%  (6.050% APR)

  Avg: 5.628%  |  vs yesterday: down 0.015%

---

Getting Started

Install

Note: On Windows, use python instead of python3. On macOS/Linux, either works.

git clone https://github.com/seang1121/Multi-Lender-Mortgage-Rate-Lookup.git
cd Multi-Lender-Mortgage-Rate-Lookup

# Create a virtual environment (recommended)
python -m venv venv

# Activate it:
# Windows (PowerShell):  venv\Scripts\Activate.ps1
# Windows (Git Bash):    source venv/Scripts/activate
# macOS / Linux:         source venv/bin/activate

pip install -r requirements.txt
python -m patchright install chromium

Configure

Set your ZIP code in config.json:

{
  "zip_code": "90210"
}

Or pass it on the command line (overrides config.json):

python mortgage_rate_report.py --zip 90210

If neither is set, defaults to 12540 (LaGrangeville, NY).

Run

# Headless mode — scrapes all 10 lenders + 2 benchmarks
python mortgage_rate_report.py

# Headed mode — visible browser, useful for debugging
python mortgage_rate_report.py --headed

# Override ZIP code
python mortgage_rate_report.py --zip 10001

---

Run It as an Agent, Bot, or Scheduled Task

The script outputs clean text to stdout. Pipe it anywhere.

Daily cron job (macOS/Linux)

0 8 * * * cd /path/to/Multi-Lender-Mortgage-Rate-Lookup && python mortgage_rate_report.py >> rates.log 2>&1

Windows Task Scheduler

schtasks /create /tn "MortgageRates" /tr "python C:\path\to\mortgage_rate_report.py" /sc daily /st 08:00

Discord bot

import subprocess

output = subprocess.run(
    ["python", "mortgage_rate_report.py"],
    capture_output=True, text=True
)
send_to_discord(channel_id, output.stdout)

Slack webhook

python mortgage_rate_report.py | curl -X POST -H 'Content-type: application/json' \
  -d "{\"text\": \"$(cat -)\"}" https://hooks.slack.com/services/YOUR/WEBHOOK/URL

AI agent integration

Your agent runs the script, reads the output, and delivers it however you want — Discord, Telegram, email, SMS. The script handles all the scraping and formatting. Your agent just schedules and delivers.

result = run_command("python mortgage_rate_report.py")
post_to_channel(result)

Rate history for analysis

90 days of rate data stored at data/mortgage_rates_history.json. Use it for trend analysis, charting, or feeding into financial models.

---

Headless vs Headed Mode

Mode	Command	When to use
Headless (default)	python mortgage_rate_report.py	Daily automation, cron jobs, servers. All 10 lenders work headless.
Headed	python mortgage_rate_report.py --headed	Debugging. Opens a visible browser so you can watch what each bank page is doing. Use this when a lender stops returning data — you'll see exactly what changed.

---

How It Works

                mortgage_rate_report.py
                        |
          asyncio.gather() — all in parallel
                        |
        +---------------+---------------+
        |               |               |
   patchright       patchright       urllib
   (stealth)       (stealth)       (CSV API)
        |               |               |
   BofA, WF,          MND          Freddie Mac
  Chase, Citi,       Index           PMMS
  Navy Fed, SoFi,
  USB, Guaranteed
  Rate, Truist,
  Mr. Cooper

All 10 lenders are scraped simultaneously with patchright — a stealth Chromium browser engine that bypasses bot detection. Each lender gets its own browser context, running in parallel batches of 4 via asyncio.gather().

Freddie Mac is fetched via direct CSV API — no browser needed.

Mortgage News Daily is fetched via plain HTML with urllib — no browser needed.

---

Tech Stack

Component	What	Why
Python 3.10+	Core language	Async support, modern syntax
patchright	Stealth Chromium	Bypasses bot detection on bank sites
asyncio	Parallelism	All lenders scraped simultaneously
Freddie Mac CSV	Benchmark API	No browser needed, direct data
JSON	Storage	90-day rolling rate history

---

Project Structure

Multi-Lender-Mortgage-Rate-Lookup/
├── mortgage_rate_report.py    # Multi-lender parallel scraper
├── config.json                # Your ZIP code
├── requirements.txt           # Python dependencies
├── CLAUDE.md                  # Project brain for AI agents
├── data/                      # Rate history (auto-created, gitignored)
└── README.md

---

Troubleshooting

patchright install fails or Chromium won't download

• Make sure you ran python -m patchright install chromium after installing requirements
• On corporate networks, proxy settings may block the download. Set HTTPS_PROXY env var if needed
• Try running as administrator/sudo if you get permission errors

SSL errors (CERTIFICATE_VERIFY_FAILED)

• Common on macOS. Run: pip install certifi and ensure your Python is using updated CA certs
• On Windows, this is rare but can happen behind corporate firewalls

All lenders return empty results

• Some lenders rate-limit or block cloud IP ranges. Try again after a few minutes
• Use --headed mode to see what the browser is actually loading
• Check your internet connection and ensure no VPN is interfering

python3 not found (Windows)

• Use python instead of python3 on Windows
• Make sure Python is in your PATH: python --version should print your version

Emoji not rendering in terminal

• The report uses emoji characters. Most modern terminals support them
• If your terminal shows garbled text, redirect output to a file: python mortgage_rate_report.py > report.txt

---

Roadmap

• MCP Server — Wrap as a Model Context Protocol server so any AI agent can call get_rates(zip_code) and get structured rate data back. Plug into Claude, ChatGPT, or any MCP-compatible platform as a live tool.
• More lenders — Community-sourced URLs and API endpoints for anti-bot protected banks
• Rate alerts — Notifications when rates drop below your target
• Historical charts — Visualize 90-day rate trends per lender
• Multi-ZIP support — Compare rates across different markets simultaneously

---

Contributing

Pull requests welcome. The fastest way to improve coverage:

• Found a working URL or API endpoint for a browser-assisted lender? Open an issue.
• Built an integration (Discord bot, Slack app, Home Assistant)? Share it.
• Have a lender we're missing? Submit a PR.

---

Disclaimer

For informational purposes only. Rates are scraped from publicly available pages and may not match actual lender quotes. Always verify directly with lenders before making financial decisions.

Related Projects

• Sports Betting MCP Server — The first MCP server for sports betting. 9 tools, live odds, AI picks, 59.6% win rate.
• March Madness Bracket Predictor — 5-model ensemble, 14-year backtest, 77% accuracy. Interactive bracket generator.
• Agent Command Center — Dashboard that auto-discovers MCP servers, agents, hooks, cron jobs, and repos.

License

MIT