Integrations·AI & ML
Give your autonomous agents a live connection to Replicate: managed, secure, and ready in minutes.
Replicate is a cloud API platform for running AI models without managing infrastructure. On Molted, your OpenClaw agents connect to Replicate through the managed MCP integration layer, with no glue code, no servers to run, and no credentials baked into containers. The agent can call models, run inference, and chain AI capabilities for Replicate, autonomously and within the guardrails you set.
Available actions
Your OpenClaw agents can call any of these Replicate actions directly through Molted's managed MCP layer. No wrappers, no boilerplate.
REPLICATE_COLLECTIONS_LIST
Tool to list all collections of models. use when you need to retrieve available model collections.
REPLICATE_CREATE_FILE
Tool to create a file by uploading content. use when you need to upload and store a file for later reference.
REPLICATE_CREATE_PREDICTION
Tool to create a prediction for a given deployment. use when you need to run model inference with specified inputs. use 'wait for' to wait until the prediction…
REPLICATE_FILES_GET
Tool to get details of a file by its id. use when you need to inspect uploaded file information before further operations.
REPLICATE_FILES_LIST
Tool to list all files created by the user or organization. use after authenticating to fetch files list.
REPLICATE_MODELS_EXAMPLES_LIST
Tool to list example predictions for a specific model. use when you want to retrieve author-provided illustrative examples after identifying the model.
REPLICATE_MODELS_GET
Tool to get details of a specific model by owner and name. use when you need model metadata (schema, urls) before running predictions.
REPLICATE_MODELS_README_GET
Tool to get the readme content for a model in markdown format. use after retrieving model details when you want to view its documentation.
How the connection works
Replicate connects with an API key you provide once; it is encrypted at rest and never shipped inside the agent's pod. Everything runs through Molted's managed MCP layer, so there is no integration code to write and nothing to keep running.
01
Your OpenClaw agent asks for Replicate when a task needs it.
02
Authorize Replicate once (API key). Credentials are encrypted at rest and scoped to what you allow.
03
It reads and writes to Replicate autonomously, inside your guardrails, with full audit and revocable access.
FAQ
Q.01
Yes. Replicate is one of 1,000+ apps available through Molted's managed MCP integration layer. Your agent requests access, you approve the connection, and it can act on Replicate straight away, with no integration code to write or maintain.
Q.02
Replicate connects with an API key you provide once; it is encrypted at rest and never shipped inside the agent's pod.
Q.03
Within the permissions you grant, an agent can call models, run inference, and chain AI capabilities. Replicate lets your agent add specialised intelligence the base model does not cover.
Q.04
Yes. Molted never bakes credentials into the agent's pod: every connection is brokered through the managed integration layer, secrets are encrypted at rest, and access stays scoped to what you approve. You can revoke Replicate access at any time.
Q.05
No. Molted is the managed operating environment for autonomous agents: compute, the MCP layer, browser automation and self-healing are all handled for you. Connecting Replicate is a configuration step, not an engineering project.
Q.06
No. Replicate is part of Molted's managed MCP integration layer at no extra per-app fee. You pay for agent compute and your plan; connecting Replicate and the other 1,000+ apps is included.
Get an agent-ready environment, the managed MCP layer with 1,000+ integrations, and direct access to our engineers. Onboarding within days.