microsoft/autogen · error · ValueError

video_path must be a local file path, not a URL.

Error message

video_path must be a local file path, not a URL.

What it means

The extract_audio tool in autogen_ext.agents.video_surfer invokes ffmpeg on a caller-supplied video_path. As an SSRF guard it rejects any path matching a URL scheme (scheme://...) with a ValueError — video must be a local file. Only http(s), file, ftp, etc. are local to the process via ffmpeg protocol handlers, which could be abused to reach internal hosts.

Source

Thrown at python/packages/autogen-ext/src/autogen_ext/agents/video_surfer/tools.py:28

    ChatCompletionClient,
    UserMessage,
)


def extract_audio(video_path: str, audio_output_path: str) -> str:
    """
    Extracts audio from a video file and saves it as an MP3 file.

    :param video_path: Path to the video file (must be a local file path, not a URL).
    :param audio_output_path: Path to save the extracted audio file (must end with .mp3).
    :return: Confirmation message with the path to the saved audio file.
    """
    import os
    import re

    # Reject URLs to prevent SSRF via ffmpeg
    if re.match(r"^[a-zA-Z][a-zA-Z0-9+\-.]*://", video_path):
        raise ValueError("video_path must be a local file path, not a URL.")

    # Enforce .mp3 extension to prevent writing arbitrary file types
    if not audio_output_path.lower().endswith(".mp3"):
        raise ValueError("audio_output_path must end with .mp3.")

    # Prevent path traversal — output must stay within the current working directory
    cwd = os.path.realpath(os.getcwd())
    output_real = os.path.realpath(audio_output_path)
    if not output_real.startswith(cwd + os.sep) and output_real != cwd:
        raise ValueError("audio_output_path must be within the current working directory.")

    (ffmpeg.input(video_path).output(audio_output_path, format="mp3").run(quiet=True, overwrite_output=True))  # type: ignore
    return f"Audio extracted and saved to {audio_output_path}."


def transcribe_audio_with_timestamps(audio_path: str) -> str:
    """
    Transcribes the audio file with timestamps using the Whisper model.

View on GitHub (pinned to 027ecf0a37)

Solutions

1. Download the video to a local file first (e.g. httpx/requests), then pass the local path to extract_audio.
2. Validate/sanitize model-provided arguments before tool execution so URLs never reach the tool.
3. In agent instructions, state that video_path must be a local file path.

Example fix

# before
extract_audio("https://cdn.example.com/clip.mp4", "/tmp/out/clip.mp3")

# after
import httpx
video_local = "/tmp/work/clip.mp4"
with httpx.Client(follow_redirects=True) as http, open(video_local, "wb") as f:
    f.write(http.get("https://cdn.example.com/clip.mp4").content)
extract_audio(video_local, "/tmp/work/clip.mp3")

import re, os

def is_local_video_path(path: str) -> bool:
    return not re.match(r"^[a-zA-Z][a-zA-Z0-9+\-.]*://", path) and os.path.exists(path)

import re

def is_local_path(p: str) -> bool:
    return isinstance(p, str) and not re.match(r"^[a-zA-Z][a-zA-Z0-9+\-.]*://", p)

When it happens

Trigger: Calling extract_audio(video_path="https://example.com/clip.mp4", ...) or any string with an <alpha><alnum+.->:// prefix (file://, ftp://, smb://, ...).

Common situations: LLM-driven VideoSurfer agents passing a scraped URL straight into the tool; user input containing a URL where a downloaded local path was expected.

Related errors

• audio_output_path must end with .mp3.
• audio_output_path must be within the current working directo
• Working directory (cwd) '{cwd}' is not valid. It must be wit
• Cannot open video file {video_path}
• Failed to capture frame at {timestamp:.2f}s

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/dc095ebc6b7527b1. Report an issue: GitHub.