Skill Detail

GStreamer Pipeline Graph Optimizer

Analyzes and optimizes GStreamer media pipelines by parsing DOT graph dumps from GST_DEBUG_DUMP_DOT_DIR, profiling element throughput via gst-stats, and suggesting queue sizing and thread pool configurations.

Media & TranscriptionOpenClaw
Media & Transcription OpenClaw Published
INSTALL WITH ANY AGENT
npx skills add agentskillexchange/skills --skill gstreamer-pipeline-graph-optimizer Copy
Works best when you want a reusable capability, not another fragile one-off prompt.
At a glance
Author
GStreamer
Last updated
Mar 24, 2026
Quick brief

The GStreamer Pipeline Graph Optimizer inspects running GStreamer pipelines to identify bottlenecks and optimize throughput. It collects pipeline topology from DOT graph dumps generated by setting GST_DEBUG_DUMP_DOT_DIR, then parses element connections, pad capabilities, and negotiated formats. The agent runs gst-stats on tracer logs to measure per-element processing time, buffer flow rates, and queue occupancy. It identifies suboptimal patterns: missing hardware acceleration (switching from x264enc to vaapih264enc or nvh264enc), unnecessary format conversions (redundant videoconvert elements), and queue starvation causing pipeline stalls. Recommendations include queue max-size-buffers tuning, multiqueue configuration for parallel branch processing, and GstTaskPool thread allocation. For live streaming pipelines, it validates latency budgets using gst-launch-1.0 –gst-debug=GST_SCHEDULING:5 and suggests leaky queue modes. Supports both playback (playbin3) and encoding (encodebin) pipeline archetypes with GstEncodingProfile optimization.