Run your Bitmovin stack
with a governed agent
Lunen agents work across Bitmovin Encoding, Observability and Stream Lab through Bitmovin's MCP server, in a chat or on a schedule. They act as the member who connected them, see only what that member's Bitmovin account can, and follow the policy you set for each tool.
Stop chasing playback issues
one dashboard at a time
The agent reads encoding logs and playback data, and runs the tests you trust it with. You decide, tool by tool, what runs on its own and what waits for a person.
Without Lunen: Finding out why last night's live encoding started late means opening it in the dashboard and reading the event log line by line.
With Lunen: Ask why it took so long. The agent reads the encoding's event log and answers in plain language: an encoder that took minutes to provision, a region fallback, a late RTMP feed.
Without Lunen: When error rate jumps, someone slices Observability by device, player version and CDN until something stands out.
With Lunen: A scheduled agent checks the night's playback data, breaks any spike down by platform and error code, and posts what it found to your team's channel.
Without Lunen: Confirming a playback problem on a real TV means setting up a Stream Lab run by hand and coming back later for the report.
With Lunen: The agent runs a Stream Lab test on the device where viewers see the problem and reads the report back, on its own or once QA approves.
Without Lunen: Giving an AI assistant Bitmovin access usually means an API key pasted into a desktop config file, no policy on what it can run, and no shared record of what it did.
With Lunen: Each member signs in to Bitmovin as themselves. Every tool has its own policy, and Lunen's audit log records who acted and what was approved.
Ask your video stack a question
Ask in plain language and get answers from live encoding and playback data, with the same identity, policies and audit log behind every call.
Video ops questions
Ask anything...
Claude Sonnet 4.6What you can do on Bitmovin
Work an agent can take on once Bitmovin is connected, whether you ask in a chat or put it on a schedule.
Encoding failure triage
Ask why an encoding failed or took too long to start. The agent reads its status and event log and explains the failure, the provisioning delay or the region fallback.
Live encoding checks
List the recent live encodings that never reached RUNNING, or summarize the last day of encoding activity by status and error.
QoE regression hunts
When startup time, rebuffering or error rate moves, the agent breaks the metric down by device, player version, CDN and country to find where it changed.
Viewer session deep dives
Paste an impression ID from a viewer complaint. The agent walks through that session's quality switches, stalls and errors.
Real-device reproduction
Run a Stream Lab test on the TVs, browsers and consoles where viewers see a problem, and read back which checks passed and which failed.
Test coverage that follows viewers
Compare the devices your viewers watch on with the ones Stream Lab tests, and flag the devices with the most playback problems and no test.
Connected through Bitmovin's MCP server
Bitmovin hosts one MCP server for all of its products. An administrator adds it to Lunen once, and each member then signs in to Bitmovin as themselves.
A policy on every Bitmovin tool
Each tool on the server has its own place in a Lunen policy. Let encoding and analytics reads run unattended and send Stream Lab test runs, which use up your monthly test executions, to a person first, or trust an agent end to end.
Bitmovin
- encoding_ask_bitmovin_encodingAllow
- observability_queryAllow
- observability_queryGroupByAllow
- observability_analyzeImpressionAllow
- general_docs_ask_bitmovin_docsAllow
- streamlab_get_test_resultsAllow
- streamlab_run_testApprove
One server for every product
Encoding, Observability, Stream Lab and the Bitmovin docs sit behind one URL, with each tool named for its product. Members see the tools their Bitmovin account can use.
No API keys to hand out
Bitmovin's server registers OAuth clients on request, so Lunen registers itself. There is no client ID, secret or API key to copy, and every member signs in with their own account.
Reads by default
The encoding and Observability tools only read. Running a Stream Lab test is the kind of action you can send to a person first.
Connected in three steps
- 1In Lunen, open Settings, then MCP Registry, and add a custom server with the URL mcp.bitmovin.com.
- 2Choose OAuth and run discovery. Lunen finds Bitmovin's sign-in and registers itself, so there is no client to set up.
- 3Each member connects Bitmovin from Connected Services and signs in with their own Bitmovin account.
Why run your Bitmovin agents in Lunen
It acts as the person
Each member signs in to Bitmovin with their own account. An agent sees the encodings, analytics and tests that account can see, and nothing more.
You decide what runs on its own
Every Bitmovin tool has its own policy. Allow the reads a trusted agent runs end to end, and set Stream Lab test runs to Approve so a reviewer sees the stream and device first.
In a chat or on a schedule
Ask about a session when a viewer complains, or put the same work on a schedule so the overnight playback check is done before the morning standup.
One record for every tool
Bitmovin calls land in the same audit log as every other tool: who acted, which agent ran, what was approved, and which model answered.
Put an agent on your video pipeline
Lunen is in early access with a small group of design partners. Tell us how your team works with Bitmovin and we'll set up a call.