What causes "GitHub API response exceeded the size limit" in 666ghj/MiroFish?

Verify nothing is intercepting the request (proxy, VPN, captive portal) — the real GitHub repo payload is far below 1 MB.

666ghj/MiroFish · error · FetchError

GitHub API response exceeded the size limit

Error message

GitHub API response exceeded the size limit

What it means

Raised by _read_response when the declared Content-Length is negative or exceeds MAX_HTTP_BYTES (1,000,000). The script deliberately refuses to buffer oversized bodies: GitHub's repo JSON is a few KB, so anything near 1 MB signals a wrong or malicious response, and the check on the header prevents allocating the read at all.

Source

Thrown at scripts/fetch_star_count.py:64

        return FetchError("GitHub API request was denied")
    if status == 404:
        return FetchError("repository metadata was not found")
    if status == 429:
        return FetchError("GitHub API rate limit was exhausted")
    if 500 <= status <= 599:
        return FetchError("GitHub API is unavailable")
    return FetchError("GitHub API request failed")


def _read_response(response: Any) -> bytes:
    raw_length = response.headers.get("Content-Length")
    if raw_length is not None:
        try:
            content_length = int(raw_length, 10)
        except (TypeError, ValueError) as exc:
            raise FetchError("GitHub API returned invalid response metadata") from exc
        if content_length < 0 or content_length > MAX_HTTP_BYTES:
            raise FetchError("GitHub API response exceeded the size limit")

    payload = response.read(MAX_HTTP_BYTES + 1)
    if len(payload) > MAX_HTTP_BYTES:
        raise FetchError("GitHub API response exceeded the size limit")
    return payload


def fetch_star_count(token: str, opener: Any | None = None) -> int:
    if not token or len(token) > 4_096 or "\r" in token or "\n" in token:
        raise FetchError("GITHUB_TOKEN is missing or invalid")

    request = urllib.request.Request(
        API_URL,
        headers={
            "Accept": "application/vnd.github+json",
            "Authorization": f"Bearer {token}",
            "User-Agent": "Repository-Star-History-Fetcher",
            "X-GitHub-Api-Version": API_VERSION,

View on GitHub (pinned to b5b53acc57)

Solutions

Verify nothing is intercepting the request (proxy, VPN, captive portal) — the real GitHub repo payload is far below 1 MB.
Confirm API_URL still targets a single repository endpoint (/repos/{owner}/{repo}) and was not repointed at a list endpoint.
In tests, set Content-Length to a realistic small value or omit it entirely (the check only runs when the header is present).

Defensive patterns

Strategy: try-catch

Try / catch

try:
    count = fetch_star_count(token)
except FetchError as exc:
    if exc.args[0] == "GitHub API response exceeded the size limit":
        # header declared an oversized body: never retry into an unknown proxy
        raise

Prevention

Keep MAX_HTTP_BYTES sized to the real payload (repo JSON is ~5 KB; 1 MB is already generous).
Do not raise the limit to make an intercepted response pass — find out what is returning megabytes for a repo lookup.
Assert in CI that direct curl of the endpoint stays under a few KB so regressions surface before the script does.

When it happens

Trigger: A Content-Length header > 1000000 or < 0 on the response from https://api.github.com/repos/666ghj/MiroFish; a proxy returning an oversized error/captive-portal page; a test double that sets an unrealistic Content-Length.

Common situations: Captive portals or proxy block pages served with huge bodies; API URL changed to point at something that returns large payloads; tests that copy headers from a different endpoint.

Related errors

AI-assisted analysis of 666ghj/MiroFish@b5b53acc57 (2026-08-14). Data as JSON: /api/errors/1def2086edc97c10. Report an issue: GitHub.