Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/clyrisai/gitresolve/llms.txt

Use this file to discover all available pages before exploring further.

The simplest way to use GitResolve is to pass a single URL or local file path as a positional argument. GitResolve automatically classifies the input, selects the right processing strategy, and prints a resolved candidate profile to your terminal.

​
Basic Usage

gitresolve <url-or-path>
The <url-or-path> argument accepts any of the input types documented below. You do not need to specify a mode flag — GitResolve infers what the input is and handles it accordingly.

​
URL Normalization

GitResolve normalizes bare domains before processing so you do not need to type the full https:// prefix every time. The following inputs are treated as-is and are not prefixed:
  • Inputs that already start with http:// or https://
  • Local file paths starting with ./, /, or ~/
  • File names ending in .pdf, .doc, .docx, or .rtf
Everything else has https:// prepended automatically: 
# These two commands resolve the same page
gitresolve github.com/janedoe
gitresolve https://github.com/janedoe

​
Supported Input Types

Input TypeExampleProcessing Strategy
Portfolio websitehttps://janedoe.devScrapes page for Git links
GitHub profilehttps://github.com/janedoeScrapes profile, discovers repos
GitLab profilehttps://gitlab.com/janedoeScrapes profile, handles /users/ routes
Bitbucket profilehttps://bitbucket.org/janedoeScrapes profile page
Repository URLhttps://github.com/janedoe/my-projectScrapes repo page, resolves owner
Remote resume PDFhttps://example.com/resume.pdfDownloads PDF, extracts text and links
Local resume PDF./resumes/janedoe.pdfReads PDF from disk, extracts text and links

​
Portfolio Website

gitresolve https://janedoe.dev
The page is fetched and scraped for any GitHub, GitLab, or Bitbucket URLs. All discovered links are classified, deduped, and organized into owned repos, contributions, and external references.

​
GitHub Profile

gitresolve https://github.com/janedoe
GitResolve detects the github.com hostname, recognizes the path as a single-segment user profile, and scrapes the profile page to extract all linked repositories.

​
GitLab Profile

gitresolve https://gitlab.com/janedoe
Both the standard route (/janedoe) and the long-form /users/janedoe route are handled correctly.

​
Bitbucket Profile

gitresolve https://bitbucket.org/janedoe

​
Repository URL

gitresolve https://github.com/janedoe/my-project
The owner username is extracted from the repo path. GitResolve scrapes the repository page and resolves the owning profile.

​
Remote Resume PDF

gitresolve https://example.com/janedoe-resume.pdf
GitResolve fetches the PDF, saves it to a temporary file, and extracts both the raw text content and any embedded hyperlink annotations. All Git URLs found are classified and returned.

​
Local Resume PDF

gitresolve ./resumes/janedoe.pdf
Local file paths are read directly from disk. The .pdf extension — or the leading ./, /, or ~/ — prevents URL normalization from prepending https://.
To process a directory of local PDF resumes, use --resumes or --resumes-dir batch mode instead of the single-URL argument. See Batch Processing for details.

​
The --type Flag

When GitResolve cannot determine the input type from the URL alone — for example, a resume hosted behind a redirect or a URL without a .pdf extension — use --type to provide a hint. 
# Force a URL to be treated as a resume PDF
gitresolve https://storage.example.com/candidate-cv --type resume

# Force a URL to be treated as a portfolio page
gitresolve https://example.com/about --type portfolio
Accepted values are portfolio and resume. The --type flag overrides the classifier but only applies to the single-URL mode — it is ignored during batch processing.

​
Classification Pipeline

When you pass a URL, GitResolve runs through the following steps before any network request is made:
1

Normalize input

The raw input string is checked for a protocol prefix, file extension, and local path markers. A https:// prefix is prepended if needed.
2

Classify input type

The classifier checks the hostname against known Git providers (github.com, gitlab.com, bitbucket.org), runs a repo-URL parser to distinguish profile URLs from repo URLs, and flags URLs with document extensions as resume inputs. Any other valid URL is classified as a portfolio.
3

Select processing strategy

Resume URLs are routed to the PDF fetcher and parser. Portfolio pages, Git profiles, and repo URLs are routed to the portfolio scraper with the appropriate browser provider.
4

Scrape or parse

For portfolio and Git URLs, the page is fetched (via fetch, Puppeteer, or Browserless) and all Git URLs are extracted from the HTML. For remote resume PDFs, the file is downloaded to a temporary path and text content and hyperlink annotations are extracted.
5

Disambiguate owner

All discovered links are analysed to determine which username appears most frequently as an owner. That username becomes the ownerProfile, and links are sorted into ownedRepos, contributions, and externalRepos.
6

Return result

A structured AggregatedResult object is printed to the terminal (or emitted as JSON if --json is set).

​
JSON Output

Add --json to suppress the ANSI terminal UI and print a raw JSON array to stdout instead. The single-URL mode still returns an array — it will contain one AggregatedResult object. 
gitresolve https://github.com/janedoe --json

[
  {
    "candidateUsername": "janedoe",
    "sources": ["https://github.com/janedoe"],
    "sourceTypes": ["git_profile"],
    "ownerProfile": {
      "url": "https://github.com/janedoe",
      "provider": "github",
      "type": "profile",
      "username": "janedoe"
    },
    "confidence": "high",
    "ownedRepos": [
      {
        "url": "https://github.com/janedoe/my-project",
        "provider": "github",
        "type": "repo",
        "username": "janedoe",
        "repo": "my-project"
      }
    ],
    "contributions": [],
    "externalRepos": [],
    "allLinks": [],
    "warnings": []
  }
]
Pipe --json output directly to jq for quick field extraction:
gitresolve https://github.com/janedoe --json | jq '.[0].ownerProfile'

​
Unresolvable Input Types

Some input types are recognized by the classifier but cannot be resolved to a Git profile. In these cases the result will always have confidence: "none", ownerProfile: null, and a human-readable explanation in the warnings array.

​
LinkedIn URLs

LinkedIn URLs are classified with an InputType of linkedin. GitResolve does not scrape LinkedIn pages — passing a LinkedIn URL returns a result with confidence: "none" and a warning explaining that LinkedIn cannot be resolved. Use a portfolio URL or a direct Git profile link instead.
LinkedIn classification exists to give you a clear warning rather than a silent failure. If a candidate’s only provided link is LinkedIn, you will see the URL echoed back with an explanatory message in warnings.

​
Unknown Input Types

If the classifier cannot determine a URL’s type — for example, a short-link redirect or an unrecognized hostname that is neither a known Git provider nor a portfolio-style domain — the input is classified as unknown. Like linkedin, an unknown input type cannot be resolved: the result will have confidence: "none" and a warning of the form "Input classified as 'unknown' — cannot resolve". 
# Example: an unrecognized short-link
gitresolve https://bit.ly/candidate-profile
# → confidence: "none", warnings: ["Input classified as 'unknown' — cannot resolve"]

Build docs developers (and LLMs) love

Get started for freeTalk to us