Gradient Inference
v0.1.3Community skill (unofficial) for DigitalOcean Gradient AI Serverless Inference. Discover available models and pricing, run chat completions or the Responses...
⭐ 0· 675·0 current·0 all-time
bySimon DeLorean@simondelorean
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the implemented functionality. Required binary (python3) and primary env var (GRADIENT_API_KEY) are appropriate for calling the Gradient inference endpoints. Declared pip deps (requests, beautifulsoup4) align with the included scripts.
Instruction Scope
SKILL.md and the scripts only instruct the agent to call the documented Gradient endpoints (inference.do-ai.run) or to scrape the public pricing docs. The scripts access only the declared env var (GRADIENT_API_KEY) for inference calls. The pricing script scrapes a public docs page and falls back to a bundled snapshot; it writes a cache to /tmp and reads the bundled JSON from the repo.
Install Mechanism
There is no remote install/download step; this is an instruction/code-only skill. The pip packages referenced are standard (requests, beautifulsoup4). No installer downloads arbitrary archives or executes code from unknown URLs.
Credentials
Only one sensitive credential is required: GRADIENT_API_KEY (the Model Access Key) — which is exactly what an inference client needs. The scripts do not request unrelated secrets or config paths.
Persistence & Privilege
always:false (normal). The skill writes a cache file to /tmp and may save generated images to the current working directory (the image saver includes a path-traversal check). It does not modify other skills or system-wide agent settings.
Assessment
This is an unofficial community client for DigitalOcean Gradient. It appears coherent and limited to its stated purpose, but review these before installing: (1) Only provide a Gradient Model Access Key — do not reuse higher-privilege or unrelated credentials. Consider creating a limited model-access key in the DigitalOcean console. (2) The pricing tool scrapes the public docs and writes a cache to /tmp; if you run this in a shared environment be aware of the cache file. (3) The code depends on requests and beautifulsoup4 — install vetted packages from PyPI. (4) Because this is community-maintained (not official), audit the code if you plan to run it in a sensitive environment or grant it autonomous invocation. If you only need basic inference, running the included scripts locally with a dedicated model-access key is a low-risk option.Like a lobster shell, security has layers — review code before you run it.
chat-completionsvk974h1s0bjb7m83zxm5jh9v75n8198ckdigitaloceanvk974h1s0bjb7m83zxm5jh9v75n8198ckgradient-aivk974h1s0bjb7m83zxm5jh9v75n8198ckimage-generationvk974h1s0bjb7m83zxm5jh9v75n8198ckinferencingvk974h1s0bjb7m83zxm5jh9v75n8198cklatestvk974h1s0bjb7m83zxm5jh9v75n8198ckllmvk974h1s0bjb7m83zxm5jh9v75n8198ck
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧠 Clawdis
Binspython3
EnvGRADIENT_API_KEY
Primary envGRADIENT_API_KEY