# Transloadit MCP > Transloadit connects your AI agent to a full media processing backend. It lets you encode videos, resize images, and manage cloud files using natural conversation instead of complex web dashboards. You can build job pipelines via JSON configuration, monitor live status, track costs per month, or restart failed encoding cycles—all through simple tool calls. ## Overview - **Category:** developer-tools - **Price:** Free - **Tags:** media-encoding, file-processing, video-transcoding, image-resizing, automation, cloud-storage ## Description Transloadit hooks up your AI agent to a full media processing backend. Forget wading through complicated dashboards just to get a video encoded or some images resized; you manage all that heavy lifting using natural conversation and structured JSON payloads. This server lets you build entire job pipelines, check costs, and restart failed cycles—all without ever touching a web UI. Your agent can use this toolset to treat Transloadit like an extension of its own memory, letting it process media assets the way you'd talk to a developer teammate about a workflow. It handles everything from initial job creation to billing audits and permanent template cleanup. **Starting Media Jobs (Assemblies)** You start complex file work by passing a JSON steps configuration to `create_assembly`, which triggers an automated processing job. This single call can initiate massive tasks, whether you're encoding high-res video or running multiple image transformations. You don't just run it and forget it; you get immediate access to the history of these runs via `list_assemblies`, which pulls a record of every media assembly your account has tackled. **Monitoring and Managing Jobs** If that job is running, you need to know what’s going down. You use `get_assembly_details` to pull the live status, final output parameters, and results for any specific job ID—whether it's still churning or done for good. If a job goes rogue or hits an unexpected snag, you can immediately issue `cancel_assembly`; that action is definitive and stops whatever’s running. And if something fails midway through? No sweat. You call `replay_assembly` to re-run the whole assembly using its previous steps, no matter what its original completion status was. **Building and Managing Reusable Workflows (Templates)** The real power here is building blueprints. Instead of retyping the same encoding sequence every time, you use `create_processing_template` to save a JSON blueprint name and its associated step set. This lets your agent store an exact workflow it can reuse later for different media files. You can check what templates are already saved using `list_templates`, or grab all the specific configuration details of an existing one with `get_template_details`. When you're done with a template—say, that client project is over and you never need those steps again—you use `delete_template` to permanently wipe it out. Remember, this action can't be undone. **Billing and Auditing** Keeping tabs on costs is non-negotiable. You pull your file processing usage metrics and associated costs for any specific calendar month by calling `get_billing_usage`, which requires you to pass the date in YYYY-MM format. It keeps your operations transparent. Overall, this toolset gives your AI agent full control over media pipelines: it lets it kick off jobs (`create_assembly`), manage reusable blueprints (`create_processing_template`, `list_templates`), check what’s happening with running tasks (`get_assembly_details`), and keep a sharp eye on the books (`get_billing_usage`). ## Tools ### cancel_assembly Immediately stops a running media processing assembly. This action is final. ### create_assembly Starts an automated file processing job by accepting a JSON steps configuration. ### create_processing_template Creates a reusable JSON blueprint for encoding or resizing using a provided name and step set. ### delete_template Permanently removes a saved processing template. This action cannot be undone. ### get_assembly_details Retrieves the current status, outputs, and results for a specific media assembly job ID. ### get_billing_usage Pulls your file processing usage metrics and costs for a specified month (YYYY-MM). ### get_template_details Retrieves the full configuration details of an existing, named processing template. ### list_assemblies Lists all recent media assemblies that have been run in your account. ### list_templates Retrieves a list of every saved processing template blueprint available to you. ### replay_assembly Re-runs an assembly job using the same steps, regardless of its prior completion status. ## Prompt Examples **Prompt:** ``` Fetch the billing usage details for January 2026. ``` **Response:** ``` I've pulled the billing usage for 2026-01. Total processing measured: 14.5 GB out of included 100 GB tier. Accrued overage cost: $0.00. ``` **Prompt:** ``` Cancel the running assembly calculation currently at ID b13a4x2. ``` **Response:** ``` Assembly b13a4x2 cancellation confirmed. Processing nodes have been immediately stopped. ``` ## Capabilities ### Run Media Jobs You pass a JSON steps configuration to create an assembly, triggering complex encoding or resizing tasks. ### Check Job Status and Results Calling `get_assembly_details` retrieves the live status, final output parameters, and results for any completed or running job ID. ### Build Reusable Workflows (Templates) `create_processing_template` lets you save a JSON blueprint so your agent can reuse the exact same encoding steps later. ### Manage and List Templates You use `list_templates` to see all saved blueprints, or `delete_template` to permanently remove an outdated one. ### Track Usage Costs `get_billing_usage` lets you pull your file processing costs and usage metrics for a specific calendar month. ### Abort or Retry Jobs You can use `cancel_assembly` to stop runaway jobs, or `replay_assembly` to re-run an assembly that failed midway. ## Use Cases ### Mass Asset Deployment A marketing team needs to resize 5,000 images and encode them into three different video formats. Instead of manually running five separate jobs in the web UI, they prompt their agent: 'Create an assembly using Template X.' The agent calls `create_assembly`, runs the job, and then uses `get_assembly_details` repeatedly until all 5,000 assets are marked complete. ### Debugging a Corrupted Batch A content engineer uploads a large batch of videos, but half fail. Instead of guessing which parameters were wrong, they ask the agent to check `list_assemblies` for IDs, then use `get_template_details` on the original template to verify steps, and finally call `replay_assembly` on the specific failed ID. ### Audit Operational Costs The technical lead needs to prove that Q1 encoding costs were under budget. They don't wait for accounting; they prompt their agent: 'Get the billing usage for 2026-03.' The agent calls `get_billing_usage`, providing an immediate, auditable cost report. ### Cleaning Up Old Workflows A team pivots their media strategy and needs to retire old video formats. They first use `list_templates` to see what's available, then confirm the exact steps using `get_template_details`, and finally call `delete_template` on the obsolete blueprint. ## Benefits - Real-time progress updates: You don't have to guess if a job is stuck. Use `get_assembly_details` to pull the exact completion status or output parameters for any running task, eliminating dashboard guesswork. - Build and reuse complex workflows: Instead of writing out encoding steps every time, use `create_processing_template`. Your agent can save these blueprints and recall them instantly via `list_templates` or `get_template_details`. - Cost visibility on demand: Stop waiting for monthly reports. Call `get_billing_usage` with a specific YYYY-MM to see your operational costs right now, which is critical for tech leads auditing expenses. - Immediate failure recovery: If an assembly hangs or fails halfway through, don't panic and restart manually. Use `replay_assembly` or `cancel_assembly` to manage the job state precisely. - Native JSON orchestration: Your AI agent treats your entire workflow like a single code block. You pass structured JSON steps directly via `create_assembly`, moving beyond simple button clicks. ## How It Works The bottom line is: your AI client speaks API commands; Transloadit runs the media work. 1. Attach the Transloadit component and provide your unique Auth Key and Secret credentials. 2. The AI agent interprets a request (e.g., 'Encode this video to 1080p'). 3. The agent calls the appropriate tool (`create_assembly`), passing the necessary JSON payload, and you get back a job ID for monitoring. ## Frequently Asked Questions **How do I check if an assembly job finished successfully using get_assembly_details?** You pass the specific Assembly ID to `get_assembly_details`. The response object will contain a 'status' field. You look for 'COMPLETED' or examine the 'output_parameters' section to confirm success. **What is the difference between creating an assembly and using create_processing_template?** You use `create_processing_template` when you want to save a reusable workflow blueprint. You call `create_assembly` only when you are ready to actually run the job, referencing that saved template. **Can I check my billing costs using get_billing_usage?** Yes. This tool requires you to pass the month in YYYY-MM format (e.g., 2026-01). It returns your total processed usage and any associated overage fees for that specific time period. **If a job fails, should I use cancel_assembly or replay_assembly?** Use `cancel_assembly` if the job is currently running and needs to be stopped immediately. Use `replay_assembly` only if the job completed but failed its final checks and you need it to run through the steps one more time. **What format must I provide when using the `create_processing_template` tool?** You need two things: a name and the full steps JSON defining the pipeline. The process requires this structured data to build reusable architectures that dictate file encodings or resizing jobs. **How do I ensure my initial request for `create_assembly` is correct?** You must provide a detailed JSON object defining every step in your pipeline. This structure dictates the encoding, resizing parameters, and output formats for the automated file processing job. **Where can I see all my saved pipelines using `list_templates`?** The `list_templates` tool pulls every processing template you've saved to your account. This lets you audit and select existing architectures without having to reconstruct the full configuration JSON. **What is the risk of using the `delete_template` command?** Running `delete_template` permanently removes that processing template from your account; this action is irreversible. Always use `get_template_details` first to verify you want to delete it. **Where do I find my Auth Key and Auth Secret credentials?** Navigate directly directly to the Transloadit web frontend. Open your Account settings. On the sidebar, click on **Credentials**. You will spot your public `Auth Key` directly exposed, and right next to it the option to reveal your `Auth Secret`. Both strings are required parameters here. **Does `cancel_assembly` action cost money if triggered early?** Yes, absolutely. Halting an assembly stops Transloadit servers from expending gigabytes processing irrelevant files immediately. Consequently, acting rapidly prevents unnecessary overuse charges and quotas from draining. Use it when observing logical JSON mistakes. **Do I need Signature Authentication for Transloadit?** Yes. For production environments, it is strongly recommended to use Signature Authentication to prevent unauthorized use. Your Transloadit MCP integration handles this internally using the provided Auth Secret to sign requests securely.