Replicate

Replicate ML platform — run AI models, manage predictions, browse collections, search models. ML inference CLI.

Security Scan

This skill is a small, self-contained CLI wrapper for the Replicate API that only needs a Replicate API token and behaves consistently with its description.

Purpose & Capability

Name, description, and code all implement a Replicate API CLI (model search, predictions, collections, account). The single required credential (REPLICATE_API_TOKEN) is appropriate for the stated functionality.

Instruction Scope

Runtime instructions simply invoke the included Python script. The script does attempt to fallback to a .env file at WORKSPACE or ~/.openclaw/workspace/.env to find REPLICATE_API_TOKEN — this file-read behavior is not declared in the SKILL.md metadata (which lists no config paths). The fallback only extracts REPLICATE_API_TOKEN, but users should be aware the script will look for that .env file.

Install Mechanism

No installation spec — instruction-only plus a small stdlib Python script. No third-party downloads or package installs are required.

Credentials

Only REPLICATE_API_TOKEN is required (declared as the primary credential), which matches the skill's purpose. The script also reads the WORKSPACE environment variable to locate a .env file if the token is not in the environment; while this is a convenience, it is not declared and could cause the token to be read from a workspace file.

Persistence & Privilege

Skill does not request always: true and does not modify other skills or system-wide settings. It does not persist new credentials or install services.

Assessment

This is a small, coherent CLI for the Replicate API. Before installing: (1) verify the author/homepage if you require a trusted source; (2) only provide a REPLICATE_API_TOKEN with the minimum needed scope; (3) be aware the script will attempt to read WORKSPACE or ~/.openclaw/workspace/.env to find the token if the environment variable is unset — avoid storing other sensitive secrets in that file or make it accessible only to you; (4) the run command accepts a --webhook URL you might supply — only use webhook endpoints you control/trust to avoid sending prediction outputs to unknown servers; and (5) if you want extra assurance, review the included scripts/replicate.py contents yourself (it uses plain HTTPS requests to api.replicate.com and has no obfuscated code).

Run ML models via API — predictions, model search, and collections.

Features

• Models — browse and search
• Predictions — create, get, cancel
• Collections — curated models
• Hardware — available GPUs

Requirements

Variable	Required	Description
REPLICATE_API_TOKEN	✅	API key/token for Replicate

Quick Start

python3 {baseDir}/scripts/replicate.py search "text to image"
python3 {baseDir}/scripts/replicate.py run <version> '{"prompt":"a cat"}'
python3 {baseDir}/scripts/replicate.py predictions
python3 {baseDir}/scripts/replicate.py me

Credits

Built by M. Abidi | agxntsix.ai YouTube | GitHub Part of the AgxntSix Skill Suite for OpenClaw agents.

📅 Need help setting up OpenClaw for your business? Book a free consultation