Adaptive Stream Check
Guest access is available after setup. Full account linking verified by PowMCP: Claude only.
Use in your browser
Check one public HLS or DASH stream for packaging, rendition ladder, and delivery problems, with ffprobe evidence from real segments. Compare two deployments to see which one is healthier.
Guest includes 50 lifetime app requests. Create a free account for 100 app requests each calendar month. One allowance across every app, in your browser and your agent.One allowance across every app, in your browser and your agent. Manage usage
Guest access is available after setup. Full account linking verified by PowMCP: Claude only.
Connect this app on its own. Add other PowMCP apps whenever your agent needs another job done.
Open the apps workspace
Choose your agent. Each button opens a new tab with only this app's endpoint ready to add.
Copy one ready-to-send prompt for any agent that supports remote MCP connections.
For clients that require manual configuration, use this app-only endpoint, its JSON entry, or one terminal command.
ChatGPT: enable Developer mode in Settings → Security and login, then add this MCP endpoint from the Plugins page. Availability depends on your account and workspace policy. PowMCP has not yet verified ChatGPT account linking. OpenAI setup guide
https://powmcp.com/adaptive-stream-health/mcp {
"mcpServers": {
"powmcp-adaptive-stream-health": {
"type": "http",
"url": "https://powmcp.com/adaptive-stream-health/mcp"
}
}
}Terminal agents add this app with one command:
claude mcp add --transport http powmcp-adaptive-stream-health https://powmcp.com/adaptive-stream-health/mcpcodex mcp add powmcp-adaptive-stream-health --url https://powmcp.com/adaptive-stream-health/mcpgemini mcp add --transport http powmcp-adaptive-stream-health https://powmcp.com/adaptive-stream-health/mcpManage, disable, or remove this connection in your agent's own MCP settings. PowMCP does not label an external connection as installed without confirmation from that client.
Proof
Fetches a public HLS or DASH manifest, walks the rendition ladder, and runs ffprobe over real downloaded segments to prove the declared codecs, resolution, and frame rate match the bytes a player would actually receive
Reads a public HLS or DASH deployment the way a player would: parses the master and media playlists or the MPD, inventories up to ten renditions with their declared bandwidth, resolution, frame rate, and codecs, then downloads real init and media segments through an SSRF-guarded, byte-capped fetcher and runs ffprobe on them to check the declarations against the measured bytes. Live playlists are refetched once to prove the media sequence or publishTime actually advances, and everything cut short by a cap comes back flagged rather than silently dropped.
Request cost1 completed result
Guests get 50 lifetime app requests. Free accounts get 100 each calendar month.
stream_check
Fetches one public HLS (.m3u8) or DASH (.mpd) manifest and works down it the way a player would: master and media playlist or MPD parsing, up to ten renditions of the ladder with their declared bandwidth, resolution, frame rate and codecs, HTTP delivery evidence (redirect hops, playlist content types, cache headers, byte-range 206 behavior, per-segment fetch times), and ffprobe over real downloaded init and media segments so declared codecs, resolution and frame rate can be checked against the measured bytes. Live playlists are refetched once to prove the media sequence or publishTime advances. Returns pass/fail, severity-ranked issues, the rendition inventory, sampled-segment evidence, and explicit truncation flags. Use it for stream health, packaging, buffering, codec, and CDN delivery questions. It measures packaging and delivery, never the player or the viewer's own bandwidth. A check against a healthy CDN measures about a second; a large ladder on a slow origin takes longer and every call is hard-bounded at 110 seconds.
stream_compare
Validates exactly two public HLS or DASH manifests inside one call that shares a single rendition cap, one segment of sampling depth, one 64 MB byte budget and one deadline, then lays their format, live/VOD classification, error and warning counts, rendition ladder, top resolution, declared codecs, manifest fetch time and median segment fetch time side by side. Names the healthier deployment strictly by error count and then warning count, and reports no winner when those tie; the live-advance refetch is skipped so neither stream is measured on a different clock. A comparison against healthy CDNs measures about a second; large ladders on slow origins take longer and every call is hard-bounded at 110 seconds.
Other apps for the jobs next to Adaptive Stream Check.
Scope and boundaries for Adaptive Stream Check.
Check the public Mux HLS master playlist with one media segment sampled per rendition and live refresh enabled, or compare that HLS deployment with a DASH MPD under the same sampling depth.
Start with pass and issueCounts: pass means no error-severity issue was found, not that every rendition or segment was measured. Fix errors first, then review warnings; treat truncation or unmeasured variants or samples as incomplete. In comparisons, winner means fewer errors, then fewer warnings, and is absent on a tie.
This measures public manifest packaging, sampled media bytes, and CDN delivery at check time. It does not exercise a playback application or represent an individual viewer's bandwidth, and unsampled renditions or segments may contain problems the bounded check does not see.