abhigyanpatwari/GitNexus · error · SandboxError

transcript artifact size does not match its results row: {pa

Error message

transcript artifact size does not match its results row: {path}

What it means

Raised by `_bound_transcript_artifact` when the transcript file's actual byte size does not equal the `bytes` field recorded in its evidence metadata. The size is pinned in results.jsonl and checked against the file on disk.

Source

Thrown at eval/workflow_bench/evolve.py:398

            seen_paths.add(normalized)
        artifacts_by_row.append(artifacts)
    return artifacts_by_row


def _bound_transcript_artifact(root: Path, metadata: Any) -> str:
    relative, expected_digest, expected_size = _transcript_artifact_metadata(metadata)

    path = _results_artifact_path(root, relative, transcript=True)
    try:
        before = path.lstat()
    except OSError as exc:
        raise SandboxError(f"transcript artifact is unavailable: {path}: {exc}") from exc
    if stat.S_ISLNK(before.st_mode) or not stat.S_ISREG(before.st_mode):
        raise SandboxError(f"transcript artifact must be a regular non-symlink file: {path}")
    if stat.S_IMODE(before.st_mode) & 0o077:
        raise SandboxError(f"transcript artifact must be owner-only: {path}")
    if before.st_size != expected_size:
        raise SandboxError(f"transcript artifact size does not match its results row: {path}")

    descriptor = os.open(path, os.O_RDONLY | getattr(os, "O_NOFOLLOW", 0))
    try:
        opened = os.fstat(descriptor)
        if not stat.S_ISREG(opened.st_mode) or opened.st_dev != before.st_dev or opened.st_ino != before.st_ino:
            raise SandboxError(f"transcript artifact changed while opening: {path}")
        digest = hashlib.sha256()
        content = bytearray()
        while chunk := os.read(descriptor, 64 * 1024):
            digest.update(chunk)
            content.extend(chunk)
            if len(content) > MAX_EVIDENCE_FILE_BYTES:
                del content[: len(content) - MAX_EVIDENCE_FILE_BYTES]
        after = os.fstat(descriptor)
        if (opened.st_size, opened.st_mtime_ns) != (after.st_size, after.st_mtime_ns):
            raise SandboxError(f"transcript artifact changed while reading: {path}")
    finally:
        os.close(descriptor)

View on GitHub (pinned to d540b00184)

Solutions

1. Recompute each transcript's size and compare to results.jsonl's `bytes` field.
2. Regenerate the whole results dir so results.jsonl and transcripts/ are consistent.
3. If only one file drifted, restore it from a trusted copy with the exact bytes.

mismatches = []
for r in evidence:
    for a in r.get('transcript_artifacts', []):
        actual = (results_dir / a['path']).stat().st_size
        if actual != a['bytes']:
            mismatches.append((a['path'], a['bytes'], actual))
if mismatches:
    raise ValueError(f'size mismatches (expected, actual): {mismatches}')

When it happens

Trigger: `proposer_evidence_entries` reads a results dir whose transcript file was truncated, appended to, or replaced with a different-sized copy after results.jsonl was written.

Common situations: A partial file copy, a log-rotation tool truncating transcripts, or a results dir from a re-run whose results.jsonl still references the old sizes.

Related errors

• transcript_artifacts must be a list
• transcript_artifacts exceeds the per-row session limit of {M
• transcript_artifacts exceeds the global evidence limit of {M
• duplicate transcript artifact path: {normalized}
• transcript artifact is unavailable: {path}: {exc}

AI-assisted analysis of abhigyanpatwari/GitNexus@d540b00184 (2026-08-12). Data as JSON: /api/errors/78a93679eb28e62b. Report an issue: GitHub.