footprint

Check what an email, password, domain, or username exposes: public breach data, leak indexes, public profiles. Then score it and say what to fix.

Works with no API keys. There's a CLI and a local web GUI.

pip install -e .
footprint gui                      # web interface
footprint email you@example.com    # or the CLI
footprint                          # or an interactive menu

Authorized use only

For checking your own exposure, or assessments you're authorized to do. Using it to profile or target people isn't supported and may be illegal.

What it checks

		Needs a key?
Email	Breaches, pastes, leak records, dark-web hits, reputation, and which sites the address is registered on, merged into one timeline	no
Password	Whether it's in breach corpora, via k-anonymity	no
Domain (proof of concept)	SPF/DMARC/DKIM/DNSSEC/MTA-STS posture, subdomains, dangling-CNAME takeover checks	no
Username	Presence across ~700 sites, scored for cross-site linkability	no
Bulk	Any of the above, one subject per line	no
Watch	Re-scan a list and alert only on what's new	no

API keys (HIBP, DeHashed, IntelX, Hunter) only add more; nothing needs them. Install footprint[holehe] for keyless registered-account discovery.

Exposure score

Every result gets a 0-100 score, worked out locally from what the sources returned. No extra request, no key. It's a plain sum of the reasons listed underneath it, so the numbers always add up:

Exposure score  ████████████████░░░░  78/100 - High
  ▲ +24  appears in 210 known breaches
  ▲ +26  critical data exposed: Auth tokens, Credit cards, Passwords
  ▲ +12  a breach occurred within the last year
  ▲ + 7  15 registered account(s) discoverable from this address

Bands are 0-19 Minimal, 20-39 Low, 40-59 Moderate, 60-79 High, 80+ Critical. The total caps at 100.

Email

Finding	Points
Known breaches	6 for the first, +3 each after, max 24
Critical data leaked	what the worst class is worth, +4 each after, max 44
Identity data leaked	what the worst class is worth, +2 each after, max 22
Newest breach under a year old	12
Newest breach 1-3 years old	6
Leak records (DeHashed)	6 for the first, +2 each after, max 14
Dark-web index hits (IntelX)	4 for the first, +2 each after, max 10
Public pastes	3, plus 1 each, max 8
Registered accounts	2, plus 1 per 3 accounts, max 8
Reputation flags (EmailRep)	4 each

The two "data leaked" rows are what separate a bad breach from a boring one. Each breach lists the categories of data it exposed, and each category carries its own weight, because they are not remotely equal:

• Critical, 12 to 32 points: passwords and hashes and crypto seed phrases at the top, then private keys, social security numbers, bank and card details, passports and government IDs, auth tokens, PINs, security questions, tax records.
• Identity, 5 to 14 points: health and sexual orientation at the top, then religion, politics and ethnicity, date of birth, private messages, home address, geolocation, phone, finances, employer, vehicle and device identifiers.

The worst class sets the level and the rest adds on top. Everything else (email addresses, usernames, gender, device details) scores nothing.

That ordering is the point. One breach that leaked your password scores higher than ten that leaked nothing but your email address, and a leaked password alone lands in Moderate rather than reading as fine.

Domain

Finding	Points
No DMARC record	16
No SPF record, or SPF ending in +all	14
DMARC p=none	9
DMARC p=quarantine	3
No DNSSEC	5
No MTA-STS policy	4
MX present but no DKIM at the common selectors	4
Dangling subdomain	12 each, max 30
Over 50 subdomains public in certificate transparency	4
Breaches affecting the domain	4 for the first, +2 each after, max 12

Username

This one measures linkability, not compromise. Nothing here means an account was breached; it's about one handle tying your profiles together.

Finding	Points
Handle found on public sites	4, plus 2 per site, max 30
Found on a sensitive site	10 each, max 25
Found on two or more identity-linked sites	5 each, max 20

Which sites count comes from the category WhatsMyName gives each one: NSFW, dating, health and political are sensitive; finance and business are identity-linked.

Advice

The same categories pick the remediation steps, which is why they're specific: leaked security questions say to replace them with random strings, a leaked phone number says to set a carrier port-out PIN.

The advice is a proof of concept and may be inaccurate.

Bulk

One subject per line, plain text.

footprint batch emails.txt                    # detects the kind
footprint batch domains.txt
footprint batch passwords.txt --kind password # passwords must be explicit
footprint batch emails.txt --save --out-dir reports

Passwords are never auto-detected, because guessing that a line is a secret would be unsafe. In password mode entries are labelled by line number, so you learn which one to fix without the secret ever being shown:

   95 Critical line 1  seen 65,763x · reused
    0 Minimal  line 2  not found

A password is never used as a subject or written to a report. The webhook message does name it, since a verdict you cannot match to a password is not much use; --no-notify skips that. Lookups are keyed by hash, and passwords sharing a hash prefix share a single API call. Reuse is detected locally and costs nothing.

Bulk is in the GUI too. Paste a list or load a .txt, and rows fill in as each subject finishes.

Watch

Track subjects and hear only about what's new. Everything already reported is fingerprinted locally, so a breach is announced once.

The first check sets the baseline: it records what's already there without alerting. Changes are flagged from the second check on.

footprint watch add you@example.com
footprint watch add example.com --kind domain
footprint watch check              # exits 1 if anything is new
footprint watch check --quiet      # silent unless something appears, for cron

Windows Task Scheduler:

schtasks /create /tn "footprint-watch" /sc daily /st 09:00 ^
  /tr "C:\path\to\.venv\Scripts\python.exe -m footprint watch check --quiet"

Watch is at-most-once: a finding is marked seen the moment it's reported, so if delivery fails you won't be told again. Test your webhook first.

Webhooks

Push results to Discord, or any JSON endpoint.

footprint webhook add https://discord.com/api/webhooks/...
footprint webhook test
footprint webhook list          # redacted, a webhook URL is a secret
footprint webhook disable 1     # mute without deleting

Once a webhook is enabled every check delivers to it; there's no flag to remember. Use --no-notify to skip one run, or switch the webhook off.

Discord messages are colour-coded by severity and carry the full HTML report as an attachment, since the embed is only a summary. Bulk runs attach one report per exposed subject.

Password checks name the password in the message, so you can tell which verdict is which. --no-notify skips it.

A dead webhook never fails a scan; delivery errors are warnings.

The GUI

footprint gui                    # http://127.0.0.1:8731

One hand-written page served by Python's own http.server. No Flask, no Node, no build step, no extra dependency. Sources stream their progress as they finish rather than hanging until everything returns. API keys and webhooks are editable under Settings, saved to the same config the CLI uses.

It binds loopback only, and every API call carries a random per-run token, since any site you have open could otherwise POST to localhost in the background.

Recent keeps the last 200 scans so you can reopen one without running it again, password checks included. A password check is stored under a masked subject (h*****2), so the entry tells you which password without a report or a download filename ever holding the real one. Those entries hold real exposure data, so the CLI can list and delete them:

footprint history           # what's stored
footprint history clear     # delete all of it
footprint history off       # stop recording new scans
``` A row of source chips under the input shows what will actually run, so a
missing key isn't a surprise afterwards. Long sweeps have a Stop button, and
results export as the HTML report or the raw JSON. Password checks are never
recorded.

## Exit codes

`0` nothing found · `1` exposure found · `2` source error · `3` usage error

Any finding exits 1, which for a long-lived address is always true. `--fail-on`
switches to a severity threshold instead:

```bash
footprint domain example.com --fail-on 60

Configuration

Set keys in Settings, or as environment variables, which take precedence. Config lives at %APPDATA%\footprint\config.json or ~/.config/footprint/config.json, and supports named profiles (--profile work).

export HIBP_API_KEY=...        # email/paste lookups
export DEHASHED_EMAIL=...      # DeHashed account email
export DEHASHED_API_KEY=...    # secrets are redacted
export INTELX_API_KEY=...      # metadata only
export HUNTER_API_KEY=...      # domain email pattern
export EMAILREP_API_KEY=...    # optional; raises the keyless rate limit
export FOOTPRINT_WEBHOOKS=...  # comma-separated webhook URLs

Global flags: --json, --cache / --no-cache, --profile NAME, --timeout SECONDS. Reports are HTML: --save --out PATH.

The GUI records finished scans so you can reopen them. footprint history off turns that off, and the config file keeps the setting as "history": false.

Exposure, not credentials

footprint reports which breach, when, and what categories of data were involved. It never returns or stores plaintext credentials. Sources that expose raw secrets (DeHashed, IntelX) are integrated with their values redacted, so footprint records that a password was present, not what it was.

The password check computes the SHA-1 locally and sends only the first 5 hex characters to the Pwned Passwords range API, matching the rest on your machine. The password itself never goes to any source.

It does go to your webhook, so the message says which password the verdict is about. Use --no-notify if you'd rather it didn't. Bulk runs are unchanged; those rows stay labelled by line number.

Domain checks are a proof of concept

SPF, DMARC and MX come straight from DNS and are reliable. The rest is best-effort: DKIM only probes 15 common selector names, takeover detection samples up to 40 subdomains against 19 known services, and subdomain enumeration depends on crt.sh, which is often down (footprint says so rather than reporting zero).

Takeover detection needs both a CNAME pointing at a known service and that service's "nothing here" page, so a healthy site isn't flagged for its host.

A clean result isn't proof. Verify anything that matters with a dedicated tool.

Development

pip install -r requirements.txt   # editable install + test/lint tooling
pytest -q                         # HTTP is mocked; no network calls
ruff check .

Dependencies are declared once, in pyproject.toml. requirements.txt is a one-line shim pointing at it.

Sources

XposedOrNot · holehe · Have I Been Pwned · Pwned Passwords · EmailRep · DeHashed · Intelligence X · Hunter.io · crt.sh · Cloudflare DNS-over-HTTPS · WhatsMyName · Gravatar

Changelog

See CHANGELOG.md.

License

MIT, see LICENSE.