Video Stream Ingestion

Execution Priority

Prioritize calling this skill's scripts/run.py entrypoint first. Only write custom code after the script result is unsuitable or the script cannot cover the requested task.

Input Resolution

If the user provides a video path, local file, stream URL, or uploaded video, build the input JSON for this skill instead of writing ingestion code. Use a stable default camera_id when the user does not provide one, infer source_type=local_file for file paths, and pass the path as file_path, stream_url, or raw_segment_uri as appropriate. Only ask the user for clarification when no usable video source can be inferred.

Atomic CLI

Run this skill directly with its own script. The script does not call other skill scripts and does not depend on shared src, tools, or registry modules.

python video-stream-ingestion/scripts/run.py --video <video.mp4> --camera-id <camera_id> --source-type local_file --capture-seconds 10 --config <config.json> --output <result.json>

Parameters: --video, --stream-url, --camera-id, --source-type, --capture-seconds, --started-at, --config, --output.

Workflow

1. Read camera id, source type, stream URL or local file, and capture duration.
2. Select the source adapter for RTSP, GB28181, API, local file, or mock input.
3. Capture a short segment with timeout and retry-aware failure handling.
4. Store the raw segment and return session metadata.

Available Implementation

This skill is implemented as an atomic standalone script in its own scripts/run.py. The script contains the executable logic for this skill and must not import shared src, tools, or registry modules.

Inputs

Required:

• camera_id
• source_type: local_file, mock, rtsp, gb28181, or api

Common fields:

• stream_url or file_path
• capture_seconds
• started_at
• raw_segment_uri

Outputs

Returns raw_segment_uri, video_session_id, started_at, ended_at, duration_seconds, and optional video width/height.

Failure Modes

• MISSING_CAMERA_ID
• UNSUPPORTED_SOURCE
• SOURCE_NOT_FOUND

Constraints

Do not detect objects, sample frames, create events, or generate evidence.