Ominis-OSINT: Web Hunter

Discover online mentions of a query or username, and map the social profiles and forums that surface with them. Ominis searches public web results, reports what it finds, and gets out of the way.

MIT licensed. Maintained by AnonCatalyst with the WinterGate Intelligence Collective.

Use it on your own presence or with permission. Ominis only reads public pages, but a search can still be intrusive. You are responsible for the law, the site's terms, and the people whose data you touch.

Version 1.0

This is the full suite. It is published alongside the patched original (v0.5.0) so anyone relying on the old layout can keep it. Highlights:

• Proxies are off by default. No proxy list is downloaded or validated unless you ask for it.
• Real filters. Language, country, and date ranges are passed to each engine instead of being accepted and ignored.
• Optional results API. Set SERPBASE_API_KEY and the Google leg goes through a SerpBase-compatible JSON API; leave it unset and the scrapers run as before.
• No browser download. Username checks share one HTTPX client instead of starting a requests_html Chromium session per URL.
• Reports in four formats. txt, json, csv, and md, with file names that are safe for any query.
• A real CLI. Scriptable subcommands, exit codes, and a doctor check.

Install

git clone https://github.com/AnonCatalyst/Ominis-OSINT
cd Ominis-OSINT
python -m pip install -r requirements.txt

Install the released package from PyPI:

python -m pip install ominis-osint-suite

The old Ominis-OSINT name on PyPI is still the 2024 0.4.8 build and is no longer updated. Prefer ominis-osint-suite, the checkout above, or a release archive.

Python 3.10 or newer is required.

Use

Search

python ominis.py search "acme corp" --country US --lang en --max-results 40

Results are written to Results/ (override with --out). Choose formats with --format txt,json,csv,md.

Filters:

python ominis.py search "acme corp" --since 2025-01-01 --until 2025-12-31

Engines are duckduckgo, bing, and serpbase. The first two are scraped. serpbase needs a key:

export SERPBASE_API_KEY=...            # PowerShell: $env:SERPBASE_API_KEY="..."
python ominis.py search "acme corp" --engine duckduckgo,serpbase

With a key, the Google leg is a JSON API call. Without one, the engine is skipped with a clear message and the scraped engines still run. The API is POST https://api.serpbase.dev/google/search with the key in X-API-Key and a JSON body of q, page, gl, hl. Override the endpoint in ominis_src/config.py if you use a compatible provider. Date filters are folded into the query for this provider (after:YYYY/MM/DD before:YYYY/MM/DD), so the other engines keep their native date parameters.

Username search

python ominis.py username acme --format txt,json,md

The platform list lives in ominis_src/urls.txt, one URL template per line. --urls FILE points at your own list. Checks run concurrently (default 8), each with its own timeout. --all records non-matches too.

Proxies (opt-in)

python ominis.py proxy harvest --limit 50
python ominis.py search "acme corp" --proxy-file config/valid_proxies.txt

proxy harvest downloads candidates, validates them, and writes the working pool. Nothing else in Ominis calls it. You can also pass endpoints directly:

python ominis.py search "acme corp" --proxy 127.0.0.1:8080 --proxy socks5://127.0.0.1:1080

Bare host:port entries are HTTP proxies. http://, https://, socks5://, and socks5h:// URLs are accepted, including authenticated forms. Credentials are redacted from console output and error logs. An explicit pool is never silently bypassed: if every proxy fails, the request fails.

TLS verification is on. For a private CA set SSL_CERT_FILE (HTTPX and aiohttp both honour it). --no-verify exists for debugging and is not recommended.

Guided mode

python ominis.py with no arguments walks through a search and offers a username check at the end. Every prompt has a default; proxies default to off.

Health check

python ominis.py doctor

Try it from GitHub Actions

The Ominis demo workflow runs a search or username check on GitHub's runners and uploads the report as an artifact. No clone, no install:

1. Open Actions → Ominis demo → Run workflow.
2. Pick search, username, or both, and fill in the query.
3. Download ominis-results when it finishes.

It uses contents: read, passes inputs through environment variables rather than interpolating them into shell code, and does not use sudo or Tor.

Configuration

Variable	Meaning	Default
OMINIS_ENGINES	comma-separated engine list	duckduckgo,bing
OMINIS_LANGUAGE	language code, e.g. en	none
OMINIS_COUNTRY	country code, e.g. US	none
OMINIS_SINCE, OMINIS_UNTIL	YYYY-MM-DD bounds	none
OMINIS_MAX_RESULTS	total results to keep	50
OMINIS_PROXY_POOL	comma-separated proxy endpoints	off
SERPBASE_API_KEY	optional results API key	none
OMINIS_OUTPUT_DIR	report directory	Results
OMINIS_FORMATS	txt,json,csv,md	txt,json
OMINIS_TIMEOUT	seconds per request	15

Arguments win over environment variables. The tool never writes secrets to disk.

Output

File	Contents
Results/<query>.txt	human-readable results and summary
Results/<query>.json	summary, results, profiles
Results/<query>.csv	one row per result
Results/<query>.md	Markdown brief
Results/username-<name>.<ext>	username matches and errors

<query> is a slug: letters, digits, ., - and _ only, at most 60 characters. ../../notes becomes notes, so a query can never choose the directory a report is written to.

Development

python -m pip install -r requirements-test.txt
python -m pytest -q

The suite is offline. Transport tests run against loopback proxies and locally generated certificates; engine and report tests parse fixed fixtures. CI runs the same command on Python 3.10, 3.11, and 3.12. CodeQL runs through the repository default setup, with dependency updates from .github/dependabot.yml.

See CONTRIBUTING.md for setup, style, and pull-request expectations, and CODE_OF_CONDUCT.md for community rules.

Layout:

ominis.py              entry point (delegates to ominis_src.cli)
ominis_src/
  cli.py               subcommands and arguments
  config.py            settings, environment, proxy files
  engines.py           DuckDuckGo, Bing, SerpBase, parsers
  search.py            orchestration, de-duplication, reports
  username.py          profile checks
  report.py            safe file names and writers
  proxy_handler.py     opt-in harvesting and validation
  wizard.py            guided first run
  tools_handler.py     legacy scraping API (kept for compatibility)
  utils.py             mentions, profiles, forums
tests/                 offline tests

Security

• Reports are plain text; do not paste them into public issues if they contain personal data.
• Proxy endpoints are untrusted input. The parser rejects malformed entries and never logs credentials.
• Please report vulnerabilities privately: see SECURITY.md or use the repository's Security tab. Never open a public issue with a live key.
• No API key, token or proxy credential is written to disk by the tool.

Licence

MIT. See LICENSE. This release is not code-signed; see CODE_SIGNING.md for the SHA-256 verification steps and why there is no self-signed certificate. The WinterGate names and logos are covered by TRADEMARKS.md, not by the MIT licence.