Should you try it for your channel?
YouTube Automation Agent is worth a private evaluation if coordinating production and review is your main problem. Its documented workflow continues beyond generating an MP4: it schedules approved videos, collects analytics, and proposes lessons for an operator to approve. You still supply the credentials and make the editorial decisions.
The README says finished videos wait for fact review, media-rights confirmation, and your approval. It also says simulated or placeholder media cannot be approved or uploaded. These are documented controls, not behaviors our sandbox verified against a live channel.
What our run established
Our run on October 1, 2026 used commit 0d7eaf9. Installation returned exit code 0 in 29 seconds, and the test command returned exit code 0 in 485 seconds. The build was skipped because no build script was present.
Our sandbox's harness records command success, exit code, elapsed time, and timeout status separately. Here, neither command timed out. That establishes command completion; it does not establish that we generated a usable video, connected Google OAuth, called live media providers, or uploaded to YouTube. Those workloads were not tested.
The attached test-count field is null. A raw system-suite footer reports 46 passed and 0 failed, but that footer is not a verified count of exhaustive, unique tests across the repository. We have not reconciled complete logs or performed a new execution. The tail also includes warnings about unavailable text generation and failed channel-owner lookup, which limit what its successful checks demonstrate about live integrations.
The stored dependency measurement is 442 packages occupying 468 MB. Separately, npm's install output says it added 495 packages and audited 496. These are different recorded figures; the npm output does not replace the harness measurement or establish the audit's package coverage. The audit record has no scannedPackages field, so scan coverage is unknown.
The MrKeyoor testing methodology explains how command outcomes and dependency advisories are reported. That page supplies provenance context; its current procedure does not prove which methodology version governed this older run.
Dependency and runtime checks come first
Our recorded npm audit reported 30 advisories: 1 critical, 18 high, 9 moderate, and 2 low. These are dependency advisories, not confirmed exploitable vulnerabilities in AgentTube or a security review of the application. Review the affected dependency paths before giving an evaluation deployment valuable credentials.
The raw install log identifies Node.js v18.20.8 and warns that @google/genai@2.10.0 requires Node.js >=20.0.0. The harness labels its image lab-node:22. Preserve that discrepancy rather than assuming installation ran under Node.js 22. A successful install does not resolve the provider's engine requirement.
What you would need to verify next
The current README adds Chromium installation to its quick start and documents Google Cloud configuration, a Desktop OAuth client, and channel authorization. While the Google app remains in Testing mode, the README warns that sign-in expires after 7 days. Account setup is therefore part of operating the application, beyond installing npm dependencies.
The documented Production readiness check makes small live text and narration requests, checks channel access, and builds and decodes a temporary MP4 without uploading it. Paid image and video probes require separate opt-ins. We did not run those checks. They are a useful next step on your deployment, followed by inspecting generated output and testing a private upload yourself.
The README at commit 0d7eaf9 identifies master as v2.10.0. It documents scene-level repair, SQLite checkpoints, and reconciliation before retrying an upload whose outcome is uncertain. It also says the bundled DarkzSEO preflight works without Python and produces advisory findings. Those findings do not establish search visibility or ranking gains.
The supplied v2.4.0 release page documents the guided setup walkthrough. It does not establish that v2.4.0 is still the latest release or that it contains the features described at our tested commit. Pin the revision you intend to evaluate and read its matching documentation.
When another tool fits better
ShortGPT documents engines for shorts, longer videos, and video dubbing and translation. Consider it when those creation tasks are the work you want to customize. Its supplied documentation does not establish a replacement for AgentTube's approval, scheduling, and analytics workflow.
Remotion supports making videos programmatically with web technology. Consider it when composing video in code is the main requirement. It could fill a rendering role in a separately designed workflow, but the supplied sources do not document an AgentTube integration. Remotion also warns that some company uses require a company license.
How to read the scores
These are editorial assessments, not benchmark results. The scale is 1 for a major documented obstacle, 2 for substantial limitations, 3 for a workable option with material operator work, 4 for strong documented support with remaining verification gaps, and 5 for comprehensive support backed by direct validation in that dimension. Unknown dimensions receive no score.
Setup ease is 3 because installation completed, while OAuth, provider configuration, and the engine warning require attention. Documentation is 4 because the supplied README gives concrete setup, readiness, review, and troubleshooting instructions. Community is 3 based on visible repository attention and recent activity; those signals do not measure support quality. Maturity is 2 because the recorded advisories and unverified live production path leave substantial deployment work.

