Hermes

Fast username hunter written in Go. Hunt across 400+ sites in parallel to find accounts tied to a given username.

Features

• Wide coverage: 400+ sites.
• Fast: concurrent HTTP checks, configurable parallelism.
• Multi-target: search several usernames in one run, positional or from a file.
• Flexible output: text, JSON, CSV, or Markdown.
• Exclusion lists: skip sites inline or from a file (-exclude @list.txt).
• NSFW filter: opt out with -no-nsfw.
• Robust: retries on transient failures (5xx, 429, network errors).
• Zero deps: a single static binary.

Installation

From source (Go 1.26+)

go install github.com/anderson3x11/hermes@latest

Or clone and build:

git clone https://github.com/anderson3x11/hermes.git
cd hermes
go build .

Usage

Quick start

hermes username

Multiple usernames

hermes user1 user2 user3

From a file

hermes -file users.txt
hermes -file users.txt username
hermes -file users.txt user1 user2

One username per line, blank lines and comments are ignored, duplicates are deduplicated. You can combine -file and usernames, in this case the usernames are checked first, then the file.

Exclude sites

hermes -exclude GitHub,Twitter username
hermes -exclude @ignore.txt username
hermes -exclude GitHub,@ignore.txt,Reddit username

Output to a file

hermes -o results.txt username
hermes -csv -o results.csv username
hermes -json -o results.json -file users.txt
hermes -md -o report.md username

Flags

Flag	Default	Description
-c	20	Concurrent requests
-t	10s	Request timeout
-retries	2	Retries on transient errors (5xx, 429, network)
-v	false	Verbose — show errors and not-found
-file	—	Read additional usernames from file (one per line, # comments)
-exclude	—	Comma-separated site names to skip; @path reads names from a file
-no-nsfw	false	Skip sites tagged NSFW
-json	false	Emit results as JSON
-csv	false	Emit results as CSV (all rows: site,username,found,url,error)
-md	false	Emit results as a markdown table (hits only)
-o	—	Write output to file instead of stdout

-json, -csv, and -md are mutually exclusive.

Output formats

JSON (-json)

[
  {
    "username": "username",
    "results": [
      { "site": "GitHub", "url": "https://www.github.com/username", "found": true },
      { "site": "Keybase", "url": "https://keybase.io/username", "found": false },
      { "site": "SomeSite", "url": "https://somesite.com/username", "found": false, "error": "context deadline exceeded" }
    ],
    "summary": { "found": 108, "errored": 24, "skipped": 2, "nsfw_filtered": 0, "total": 434 }
  }
]

CSV (-csv)

One row per site check across all users, including not-found and errored rows:

site,username,found,url,error
GitHub,user,true,https://www.github.com/user,
SomeSite,user,false,https://somesite.com/user,context deadline exceeded

Markdown (-md)

A hits-only table (great for reports):

| Site | Username | URL |
|------|----------|-----|
| GitHub | user | https://www.github.com/user |

Notes

• The Go standard flag package stops parsing at the first positional argument. Always put the names last: hermes -c 40 user1 user2, not hermes user1 user2 -c 40.

Known limitations

Several high-profile sites (e.g Twitter/X, Instagram, LinkedIn, Discord...) currently return unreliable results due to anti-bot measures on the real sites and dead third-party mirrors in the inherited site list. They are silently skipped or show as errors in a normal run, so they don't clutter the output, but if one of them is consistently missing for an account you know exists, this is expected. See BROKEN_SITES.md for the per-site details and manual-check URLs.

Credits

Inspired by and seeded with the site list from the Sherlock Project.

License

MIT