How to use Aheliotech Skills

In the library, click API Keys in the top bar. Give the key a label (e.g. cursor-dev), choose a scope (Skills, Knowledge, or both) and an access level (Read-only is enough for fetching), then click Generate key. Copy the key — it is only shown once.

Keys authenticate all /library/ endpoints via Authorization: Bearer <key>.

The built-in guide skill has the ID helio-skills-library-guide. Fetch it once:

Copy the instructions field from the response. That text is your system prompt — it tells your AI the base URL, all available endpoints, how to search, and how to apply what it finds.

Paste the instructions as a persistent system prompt, project instruction, or native skill — whichever your tool supports. See the per-tool sections below for exactly where. The key and the API base URL are embedded in the text.

Alternatively, use Use → Add to Claude (or the relevant tool button) directly from the skill detail view — it copies or downloads the skill in the right format for that tool.

In any conversation, the AI can call:

It reads the returned instructions field and applies them to the task at hand — without you needing to manually copy or configure anything. You can tell it "use the best skill for writing an ITP spec" and it will search, find, and apply the right one.

If your API key has Knowledge scope, the AI can also retrieve reference documents:

Knowledge entries are factual reference material (configuration, architecture, conventions) rather than task instructions. The AI uses content from knowledge entries as background context.

The library publishes an OpenAPI 3.1 spec — no auth required:

Point any API-aware tool (Cursor, Copilot with extensions, custom agents) at this URL for auto-discovery of all endpoints, parameters, and response shapes.

Go to claude.ai → Settings → Capabilities and turn on Code execution and File creation. Team/Enterprise admins enable this org-wide under Organization settings.

Open the Helio Skills Library — API Guide skill in this library, click Use → Add to Claude. A ZIP downloads and the how-to opens. Upload it via Customize → Skills → +. This gives Claude the ability to find and load any other skill itself.

For any other skill: click Use → Add to Claude (or Download .zip) then upload via Customize → Skills → +. The ZIP contains a SKILL.md in the exact format Claude expects. .zip

Skills are available in every chat. Claude activates one when context matches, or you can ask directly: "Use my <skill name> skill" or "Find and use the best skill for writing a test plan."

On claude.ai, click Projects → New project. Projects maintain context across conversations.

Open a skill here and click Use → Copy as system prompt. In your Project, click Edit project details → Custom instructions and paste. All conversations in that project follow the skill.

Fetch the skill, then pass the instructions as the system parameter:

In your project root, paste skill instructions into CLAUDE.md. Claude Code reads this file automatically on every invocation. For team-shared instructions, commit CLAUDE.md to the repo.

Alternatively, add the guide skill as a slash command or MCP server that fetches skills from the library on demand.

Click Use → Export for VS Code. This downloads copilot-instructions.md — Copilot's native instructions format. copilot-instructions.md

Move the file to .github/copilot-instructions.md in your repo root. Commit and push. Copilot picks it up automatically in VS Code, github.com, and the CLI — no extension settings needed.

The file accepts any markdown. You can concatenate several skills' instructions into a single copilot-instructions.md — separate each with a heading so Copilot knows where one ends and another begins.

Click Use → Download .md. Rename the file to <skill-name>.prompt.md and place it in .github/prompts/ in your repo. .prompt.md

In Copilot Chat, type /<skill-name> to attach and invoke the prompt file. This keeps skills discoverable in the IDE without polluting repo-wide instructions.

Click Use → Export for Cursor. Downloads a .cursorrules file containing the skill instructions. .cursorrules

For a single skill, rename the file to .cursorrules and place it in the project root. Cursor auto-discovers it. For multiple skills, place individual .mdc files in .cursor/rules/ — each file becomes a separate rule Cursor can selectively apply.

Add the API guide skill to your rules directory. Cursor Agent mode can then call the library API to pull the right skill for any task. In chat, tell Cursor: "Before starting, check the skills library for a relevant skill."

Open Cursor Settings → Rules for AI and paste the skill instructions there. These apply across all projects — good for general coding standards you always want active.

Click Use → Export for Windsurf. Downloads a .windsurfrules file. .windsurfrules

Rename the file to .windsurfrules and place it in the root of your project. Windsurf reads it automatically and applies the instructions to Cascade and inline completions.

For rules that apply everywhere, open Windsurf Settings → Cascade → Global Rules and paste the skill instructions there.

Click Use → Export for VS Code. Downloads copilot-instructions.md in the format GitHub Copilot reads natively. copilot-instructions.md

Put the file at .github/copilot-instructions.md in your project. VS Code's Copilot extension picks it up with no configuration. This shapes both inline suggestions and Copilot Chat for that workspace.

For instructions that follow you across all projects: open VS Code Settings → search "Copilot instructions" and paste the skill text into the GitHub Copilot: Instructions field.

Click Use → Export for Warp to download a .yaml prompt file. .yaml

Move the file to ~/.warp/prompts/ (create the directory if needed). Warp scans this folder automatically.

Press the Warp AI shortcut, then browse your saved prompts. Select the skill to load its instructions as the session context. You can also reference it inline while typing a command.

For instructions that are always active, open Warp Settings → AI → Custom instructions and paste the skill text. These apply to every Warp AI session.

Click Use → Export for Amazon Q. Downloads a Markdown file formatted for Q's rules system. .md

Create the directory .amazonq/rules/ in your project root if it does not exist. Place the skill file there. Amazon Q Developer reads all files in this directory and applies them as context for that workspace.

You can add multiple files to .amazonq/rules/ — each one is a separate context source. Name them descriptively (e.g. code-review.md, spec-writing.md) so Q can pick the most relevant one.

Click Use → Copy as system prompt or Add to OpenAI. The full skill is now in your clipboard.

Go to chatgpt.com → Explore GPTs → Create (or edit an existing one). Paste the skill into the Instructions field. Users of that GPT will always have the skill active.

Copy the skill instructions and paste them as the first message in a new conversation. Tell ChatGPT: "These are your instructions for this session. Acknowledge and wait for my first task."

Fetch the skill at runtime and pass instructions as the assistant's instructions field (Assistants API) or as the first system message in the input array (Responses API). Combine with knowledge entries for richer context.

Click Use → Gemini. The skill instructions are copied to your clipboard.

On gemini.google.com, open Gem manager → New Gem. Paste the skill instructions into the Instructions field. Save and use the Gem for tasks that skill covers.

For a one-off use, start a new Gemini chat, paste the skill instructions, and say: "These are your working instructions. Ready for my first task."

In AI Studio, open a prompt, expand System instructions, and paste the skill text. Save as a preset to reuse across sessions.

Fetch the skill at request time and set it as system_instruction in your GenerateContentRequest: