Replicate Skills

1---
2name: replicate
3description: Discover, compare, and run AI models using Replicate's API
4---
5 
6## Docs
7 
8- Reference docs: https://replicate.com/docs/llms.txt
9- HTTP API schema: https://api.replicate.com/openapi.json
10- MCP server: https://mcp.replicate.com
11- Set an `Accept: text/markdown` header when requesting docs pages to get a Markdown response.
12 
13## Workflow
14 
15Here's a common workflow for using Replicate's API to run a model:
16 
171. **Choose the right model** - Search with the API or ask the user
182. **Get model metadata** - Fetch model input and output schema via API
193. **Create prediction** - POST to /v1/predictions
204. **Poll for results** - GET prediction until status is "succeeded"
215. **Return output** - Usually URLs to generated content
22 
23## Choosing models
24 
25- Use the search and collections APIs to find and compare the best models. Do not list all the models via API, as it's basically a firehose.
26- Collections are curated by Replicate staff, so they're vetted.
27- Official models are in the "official" collection.
28- Use official models because they:
29  - are always running
30  - have stable API interfaces
31  - have predictable output pricing
32  - are maintained by Replicate staff
33- If you must use a community model, be aware that it can take a long time to boot.
34- You can create always-on deployments of community models, but you pay for model uptime.
35 
36## Running models
37 
38Models take time to run. There are three ways to run a model via API and get its output:
39 
401. Create a prediction, store its id from the response, and poll until completion.
412. Set a `Prefer: wait` header when creating a prediction for a blocking synchronous response. Only recommended for very fast models.
423. Set an HTTPS webhook URL when creating a prediction, and Replicate will POST to that URL when the prediction completes.
43 
44Follow these guideliness when running models:
45 
46- Use the "POST /v1/predictions" endpoint, as it supports both official and community models.
47- Every model has its own OpenAPI schema. Always fetch and check model schemas to make sure you're setting valid inputs. Even popular models change their schemas.
48- Validate input parameters against schema constraints (minimum, maximum, enum values). Don't generate values that violate them.
49- When unsure about a parameter value, use the model's default example or omit the optional parameter.
50- Don't set optional inputs unless you have a reason to. Stick to the required inputs and let the model's defaults do the work.
51- Use HTTPS URLs for file inputs whenever possible. You can also send base64-encoded files, but they should be avoided.
52- Fire off multiple predictions concurrently. Don't wait for one to finish before starting the next.
53- Output file URLs expire after 1 hour, so back them up if you need to keep them, using a service like Cloudflare R2.
54- Webhooks are a good mechanism for receiving and storing prediction output.
55 
56 
57