GitHub · Repository contents · REST API · Checked October 25, 2026

GitHub: why the file's contents were not given

A file at a given commit is read with GET /repos/OWNER/REPO/contents/PATH?ref=COMMIT. With Accept: application/vnd.github.raw+json it answers the file itself; by default it answers JSON with the content base64-encoded. GitHub's documentation lists 200, 302, 304, 403 and 404 for it, and sets limits by size (checked October 25, 2026).

{"message": "This API returns blobs up to 1 MB in size. The requested blob is too large to fetch via the API, but you can use the Git Data API to request blobs up to 100 MB in size.", "errors": [{"resource": "Blob", "field": "data", "code": "too_large"}]}

By size

1 MB or smaller: GitHub's documentation says all features of the endpoint are supported.
Between 1 and 100 MB: only the raw and the object media types are supported. With the object type, content comes back empty and encoding is none; the raw type gives the file. In the default JSON, a public report (from 2017) quotes the refusal above, too_large, which points to the Git Data API.
Greater than 100 MB: the documentation says the endpoint is not supported. GitHub blocks files larger than 100 MiB in repositories, and warns about files over 50 MiB.
The Git Data API (GET /repos/OWNER/REPO/git/blobs/SHA) supports blobs up to 100 megabytes, and also answers raw with application/vnd.github.raw+json.

404, and answers that are not the file

404: the path is not in that commit (the file was added after it, deleted, or renamed: its old name is in the older commit), or the repository is not there for this token, which is also how GitHub answers a private repository the token cannot see.
A directory comes back as a list of entries, up to 1,000 files; for more, the documentation points to the Git Trees API.
A symlink to a normal file in the repository comes back as that file's content; otherwise as an object describing the symlink.
A submodule comes back as an object whose submodule_git_url and sha name the other repository and its commit, not as a file.
The token: in GitHub's list of permissions, this endpoint is Contents read for a fine-grained token; a classic token needs the repo scope for a private repository (see which permission to add).

What to do

Over 1 MB: ask with Accept: application/vnd.github.raw+json, or read the blob with the Git Data API.
404 for a file that is in the pull request: use the old name for the older commit of a renamed file, and nothing for the side where the file is added or deleted.
Many files, or a whole directory: read the tree with the Git Trees API.
A diff that is too large as a whole: see GitHub diff 406 "too_large".

GitHub pull requests in IntelliJ IDEA

Pull Request Inbox for GitHub, a plugin by Lesstep for JetBrains IDEs 2024.1 or later, opens each changed file in the IDE's diff viewer, reading both sides with the raw media type at the target branch's merge base and at the pull request's head: an added file has an empty left side, a deleted file an empty right side, and a renamed file is read by its old name on the left. When GitHub gives no text for a side that should be there (a 404, or a redirect to another host, as for a file in Git LFS), that side is shown empty and its heading says the text was not shown, so it does not pass for an added or deleted file; other refusals are shown in the diff window, with GitHub's words. The review inbox, diffs and conversations are free; approving, commenting and merging are Pro (US$1.50 a month, with a 30-day trial).

GitHub is a trademark of GitHub, Inc.; IntelliJ IDEA is a trademark of JetBrains s.r.o. This page is not affiliated with or endorsed by GitHub, Inc. or JetBrains.