The invalid verdict in check-if-email-exists is not a missing mailbox
Check if an email address exists without sending any email, written in Rust. Comes with a ⚙️ HTTP backend.
At a glance
- What is it?
- check-if-email-exists runs an SMTP probe from Rust, a CLI or a Docker backend and folds the whole result into one is_reachable word. That single word comes back as invalid even for an address Gmail has merely disabled, and the backend needs outbound port 25 plus Postgres and RabbitMQ to go past one address at a time.
- Who is it for?
- Use it if you already operate a host with port 25 open or you pay for SMTP proxies, and if you can keep the individual smtp flags instead of keeping only is_reachable. Do not use its verdict as proof that a person exists, and do not read the 0.9 examples as the current version.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 6 days ago.
- What is it written in?
- Mainly Rust, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- reacherhq/check-if-email-exists10,489 ★ · +536
| Repository | 2026-09-22 | 2026-11-15 |
|---|---|---|
| reacherhq/check-if-email-exists | 10k | 10.5k |
536 Stars in the last 54 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
Outbound port 25 is the part most hosts block
The whole approach depends on the machine running the check opening an SMTP conversation with the target domain's mail exchanger, so the first instruction in the project is a network requirement rather than a package step: outbound port 25 must be open. The container still has to reach port 25 on someone else's server from wherever you launched it.
docker run -p 8080:8080 reacherhq/backend:latestIf your cloud instance, office network or residential provider filters 25, the SMTP-based checks cannot work. The project is blunt about the remedy: it is possible to operate Reacher with your own IP addresses, but processing more than very small volumes will need SMTP proxy servers, and proxy25.com is the one named. The consequence for a reader is that this is not a self-contained tool. A single check from a home connection can work; a verification service or a nightly job needs SMTP proxy servers. The request body's proxy object takes host, port, username and password.
The quick start posts to v0 while the Makefile runs v1 endpoints
Copying the request from the top-level documentation sends a POST to http://localhost:8080/v0/check_email with a to_email field and an optional proxy object:
{
"to_email": "[email protected]",
"proxy": { // (optional) SOCK5 proxy to run the verification through, default is empty
"host": "my-proxy.io",
"port": 1080,
"username": "me", // (optional) Proxy username
"password": "pass" // (optional) Proxy password
}
}The Makefile in the same tree describes the backend differently. The plain run target starts the binary with a comment about single-shot verifications through the /v1/check_email endpoint, and the worker target is the one that enables the /v1/bulk endpoints. The consequence is concrete: the README and the Makefile disagree on the endpoint version, so check which path your image serves before debugging mail servers.
is_reachable says invalid for a Gmail address that is disabled
The worked example uses [email protected] and is explicit that the address is disabled by Gmail. The output still shows mx.accepts_mail as true, five gmail-smtp-in.l.google.com hostnames in mx.records, and smtp.can_connect_smtp as true. The example pairs the invalid verdict with smtp.is_disabled set to true and smtp.is_deliverable false, so is_reachable folds a disabled mailbox into the same invalid verdict it gives other undeliverable addresses. It reports one of safe, risky, invalid or unknown. The consequence for a reader is that invalid is not a synonym for nobody home. A disabled account is reported as invalid, so a signup flow that drops every address whose verdict is not safe will throw away real people with locked or forgotten accounts.
is_catch_all can make every address on a domain look deliverable
The project's own table splits the answer across fields: is_reachable, syntax.is_valid_syntax, mx.accepts_mail, misc.is_disposable, smtp.can_connect_smtp, smtp.is_deliverable, smtp.is_disabled, smtp.has_full_inbox, and a catch-all address row that fills smtp.is_catch_all. The misc block in the example also carries is_role_account and is_b2c. Read together, those flags decide how much weight a positive result can carry. On a catch-all domain the mail server answers for addresses nobody ever registered, so is_deliverable being true identifies no person. When has_full_inbox is true the mailbox plainly exists and still refuses mail. The consequence for a reader is that there is no field worth reading alone: a list built only from smtp.is_deliverable imports role accounts, catch-all domains and full inboxes as if they were personal addresses. The only defense is to keep is_catch_all, is_role_account and has_full_inbox beside the verdict instead of discarding them the moment the check returns.
The quick start pins 0.9 while the newest release is v0.11.7
Two version numbers sit in the quick start. The Cargo snippet tells you to add
[dependencies]
check-if-email-exists = "0.9"and the CLI block prints check_if_email_exists 0.9.1 above its usage line:
> $ check_if_email_exists --help
check_if_email_exists 0.9.1
Check if an email address exists without sending an email.
USAGE:
check_if_email_exists [FLAGS] [OPTIONS] [TO_EMAIL]The release list goes further: v0.11.10 on 2026-10-16 and v0.11.11 on 2026-10-30. The example a reader copies is roughly two minor series behind the newest tag, and the help text was captured at a different time than the release it is quoted from. That gap matters because the JSON output is a contract. Field names such as is_reachable, mx.accepts_mail and smtp.is_catch_all are what a parser binds to, and an older crate may not emit the same set. The consequence is a build that compiles and then quietly parses an older shape. Check which fields your chosen release actually returns before writing the parser, and take the CLI from the releases page rather than reading the 0.9.1 banner as a statement about the current binary.
The .env example sets four values and leaves port 9515 unexplained
Building the backend from source starts with configuration, and the example file tells you to copy it first:
cp .env.example .envThree settings follow: RUST_LOG=reacher=info, SQLX_OFFLINE=true and RCH_WEBDRIVER_ADDR=http://localhost:9515 (the section heading still says four). SQLX_OFFLINE=true pairs with the checked-in .sqlx/ directory at the repository root, so database queries resolve without a live Postgres connection at compile time. RCH_WEBDRIVER_ADDR points at a webdriver on port 9515, and the example file does not say which process must be listening there or how to start it. The short version of a source run is in the Makefile:
cd backend && cargo run --bin reacher_backendThe top-level tree also carries .envrc, .gitbook.yaml, a ci/ directory and a docs/ directory, so the published book is not the only source of configuration detail.
Bulk mode needs Postgres and RabbitMQ, and single-shot mode will not take the job
The backend has two run modes and the Makefile draws the line between them. Plain run starts the binary without worker mode, which the comment says enables only single-shot verifications through the check_email endpoint. The worker target exports three values before running the same binary:
run-with-worker: export RCH__WORKER__ENABLE=true
run-with-worker: export RCH__WORKER__RABBITMQ__URL=amqp://guest:guest@localhost:5672
run-with-worker: export RCH__STORAGE__POSTGRES__DB_URL=postgresql://localhost/reacherdbWorker mode is what enables the bulk endpoints, and the comment on that target says a Postgres database and a RabbitMQ instance must be running. The repository root carries rabbitmq/ and sqs/ directories alongside backend/, cli/ and core/, so queueing belongs to the design rather than being bolted on. The consequence for a reader is that the single container from the quick start is the single-address path. A batch job handed to a backend started without those two services has no queue to land in and nowhere to record its results, so the bulk route gives you nothing while the one-at-a-time call keeps answering. The credentials in those two lines are the default guest account, which is fine for a laptop and wrong for a shared host.
make changelog deletes local tags before it rewrites CHANGELOG.md
The last push was on 2026-10-30, and the newest published release is v0.11.11 from the same day, so treat the version strings quoted in the examples as frozen snapshots rather than current statements. Two Makefile targets matter if you intend to track the project over time. The changelog target strips local tags first with git tag | grep -E '(beta|backend|worker)' | xargs git tag -d, then regenerates CHANGELOG.md through conventional-changelog in the angular preset. Running it will delete local tags unless you bring them back with git fetch --tags. Beside it, the project page points at No2Bounce.com as the hosted offering, with [email protected] as the contact. The repository's license metadata says MIT and the top level carries a LICENSE.md, alongside a hosted product. Read LICENSE.md before this code ships inside a commercial product.
Editorial conclusion
Use it if you already operate a host with port 25 open or you pay for SMTP proxies, and if you can keep the individual smtp flags instead of keeping only is_reachable. Do not use its verdict as proof that a person exists, and do not read the 0.9 examples as the current version. Before shipping, check which fields your release actually emits, read LICENSE.md, and test one known-good and one known-dead address per domain type you intend to accept.
Frequently asked questions
how to check if email exists or not
Start the backend with docker run -p 8080:8080 reacherhq/backend:latest and send a POST to http://localhost:8080/v0/check_email with a to_email field. Outbound port 25 has to be open on the host for the SMTP part of the check to work at all.
how to check if email exists gmail
Gmail accounts that exist can still report invalid. For [email protected] the documented output shows is_reachable as invalid while mx.accepts_mail and smtp.can_connect_smtp are both true, and the reason given is smtp.is_disabled, because the address is disabled by Gmail.
Can I ping an email address to see if it exists?
No mail is sent to the address. The check happens by contacting the mail exchanger of the domain, which is why outbound port 25 must be open and why the CLI binary talks to mail servers directly from your own computer without any backend.
Can I check if an email is still active?
There is no single active flag. is_reachable reports one of safe, risky, invalid or unknown, and the response also carries smtp.can_connect_smtp, smtp.is_deliverable, smtp.is_disabled, smtp.has_full_inbox and smtp.is_catch_all, so an address that exists but is switched off reports invalid rather than a distinct state.
how to check if email exists python
The documented programmatic path is Rust: add check-if-email-exists to Cargo.toml and call check_email with a CheckEmailInput built from a list of addresses. The workspace members are backend, cli, core and sqs, and no Python package appears among the repository's top-level entries.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/reacherhq-check-if-email-exists)Star history chart for your README
[](https://hysenlabs.com/projects/reacherhq-check-if-email-exists)Was this useful?