DumpDork

DumpDork is a small, search-driven OSINT and reconnaissance command-line tool. It searches the web, news, images, and videos through the official Brave Search API, and repositories and code through GitHub's REST API. Queries stay in each provider's native syntax.

Install

The published release is available on PyPI. Install it with pipx (recommended for a command-line tool) or pip:

pipx install dumpdork
# or, inside a virtual environment:
pip install dumpdork

python3 -m venv .venv
.venv/bin/python -m pip install -e .

Run .venv/bin/dumpdork (or activate the environment to use dumpdork). On Windows, the executables are in .venv\Scripts\. You can also run python dumpdork.py from a checkout.

Credentials

Set BRAVE_API_KEY for Brave searches and GITHUB_TOKEN for GitHub code search and better GitHub rate limits. GitHub repository search can run without a token. Environment variables override ~/.config/dumpdork/config.yaml. Use dumpdork -w to save credentials privately and interactively. See API_SETUP_GUIDE.md. Never put real keys in the repository's example config.yaml.

Search

dumpdork 'site:*.example.com (intitle:"Swagger UI" OR inpage:"OpenAPI")' # Brave Web (default)
dumpdork -s brave --brave-type news --brave-freshness pw 'supply chain'
dumpdork -s brave --brave-type images --brave-count 100 -l 100 'public satellite imagery'
dumpdork -s brave --brave-type videos --brave-country AR 'security conference'
dumpdork --brave-paginate -l 100 'site:example.com' # explicitly fetch additional pages
dumpdork --brave-result-filter web,discussions --brave-extra-snippets 'topic'
dumpdork -s github --github-type repo 'topic:osint language:python'
dumpdork -s github --github-type code 'filename:config.php DB_PASSWORD'

-s brave and --brave-type web are the defaults. Brave fetches one page per search, even with a high --limit. To fetch more pages, use --brave-paginate -l N.

Without --limit, the result cap matches one Brave page: 20 for Web, 50 for News/Videos, or 200 for Images. GitHub defaults to 50 and keeps its own pagination behavior. --limit caps normalized results across all returned sections, not just web links.

Use --brave-count to set the number of results per page and --brave-offset to choose the starting page (0–9). Images do not support pagination. If a one-page search returns fewer results than requested and more pages are available, JSON reports incomplete: true. When pagination is enabled, DumpDork follows Brave's query.more_results_available and deduplicates overlapping results. Brave pagination.

Run dumpdork -h to see all --brave-* options. They include language, location, safe search, freshness, result filters, Goggles, and response options. Unsupported options for a search type are rejected before any API request.

Boolean flags have a --no- form, such as --no-brave-spellcheck. Prefer repeatable --brave-goggle (up to three) over the deprecated --brave-goggles-id.

Rich callback responses provide a hint; DumpDork does not call the separate Rich endpoint. For option details, see Brave's Web, News, Images, and Videos references.

Brave search operators

Put these operators directly in the query; DumpDork passes them through to Brave without translating Google dorks. The table follows Brave's Search API operator reference.

Write AND, OR, and NOT in uppercase. Brave describes operators as experimental, so complex combinations may not behave exactly like Google dorks; intext: and wildcard domains such as site:*.com are not listed in Brave's reference. Use inbody: or inpage: for text matching. --no-brave-operators disables Brave's operator interpretation for a query.

GitHub search qualifiers

Select -s github --github-type repo or -s github --github-type code. DumpDork sends the query unchanged to GitHub's REST search endpoints. These are common qualifiers from GitHub's repository search and legacy code search documentation; see those references for the full lists.

Repository search (--github-type repo):

Code search (--github-type code) uses the legacy REST API syntax, not the newer GitHub.com code-search syntax:

Numeric and date qualifiers support comparisons such as >, >=, <, and ranges such as 100..500; keep the entire query quoted in your shell so < and > are not interpreted as redirection. GitHub code search requires a token in DumpDork, indexes only default branches and files smaller than 384 KB, and generally needs a search term alongside qualifiers. Repository-only qualifiers such as stars: do not apply to code search. The newer content: qualifier is not supported by the REST code endpoint; DumpDork rejects it. --include-commits fetches three recent commits once per distinct repository.

Output and rate limits

Terminal text is the default. The original banner appears on interactive runs, not -h, and goes to stderr so stdout remains pipeline-friendly. JSON and JSONL have no banner or notices on stdout.

dumpdork 'site:example.com' --format json | jq '.results[].url'
dumpdork 'site:example.com' --format jsonl | jq -r '.url'
dumpdork 'site:example.com' -o results.json
dumpdork 'site:example.com' -o provider-pages.json --raw

With -o, DumpDork writes normalized JSON by default. Use --format jsonl for one result per line, or --raw to save redacted provider responses. Raw output also contains non-link Brave sections such as infoboxes.

Normalized JSON uses schema_version: 2. It includes the query, source, search type, status, counts, optional effective_query, and a results array. Results include a URL, title, source, kind, matched query, observation time, and optional snippet, repository, path, and metadata. Encoded URLs are preserved.

With --brave-paginate, DumpDork uses Brave's per-second rate-limit headers to pace requests. It does not retry failed requests or wait for a long quota reset. If a later page fails, collected results are kept with status partial. A one-page search avoids later-page errors, but its first request can still fail if quota is exhausted. Brave rate-limit guide.

Errors distinguish authentication, authorization, rate limits, quota, invalid queries, network failures, and malformed responses. Exit codes are 0 for success (even with no results), 1 for failure, 2 for invalid CLI usage, and 3 for partial results. Enable pipefail if you want a pipeline with jq to report DumpDork's exit code.

Queries are passed through without rewriting. Brave documents inbody: rather than Google's intext:. See the search operators.

License

DumpDork is licensed under Apache License 2.0; see LICENSE.