A repository of installable skills for Codex, Cursor, Claude Code, and compatible AI IDEs.
Each skill should be self-contained when possible. Runtime dependencies, setup scripts, and troubleshooting commands should live inside the skill directory rather than the repository root.
If your IDE supports prompt-based skill installation, use:
Set up the <skill_name> skill from https://github.com/primus-labs/skills.git, install its dependencies, and verify it with the doctor script.
If local fallback setup is needed:
git clone https://github.com/primus-labs/template-bot.git
cd template-bot/<skill-folder>
npm run setup
npm run doctor| Skill | Description | Path | Runtime | Use prompt |
|---|---|---|---|---|
primus-network-core-sdk |
Build backend/server-side Node.js programs using @primuslabs/network-core-sdk. No browser extension or user interaction required. Supports privacy ops, multiple URLs, mTLS, and DVC/zkVM workflow. |
primus-network-core-sdk/ |
Node.js (Backend) | Use the primus-network-core-sdk skill to build a program that proves <what_to_prove>. The requests are <url_list>. My wallet private key is <privatekey>. Chain: <chain>. |
primus-network-js-sdk-dapp |
Scaffold a DApp using @primuslabs/network-js-sdk (decentralized zkTLS). Requires wallet + gas for on-chain task submission. Default to Base Sepolia for dev, Base mainnet for production. |
primus-network-js-sdk-dapp/ |
Node.js + Vite | Use the primus-network-js-sdk-dapp skill to scaffold a DApp that proves <what_to_prove>. The template ID is <template_id>. |
primus-zktls-js-sdk-dapp |
Build frontend DApps using @primuslabs/zktls-js-sdk (enterprise zkTLS). Supports test mode (frontend signing) and production mode (backend signing). No wallet/gas required for attestation generation. |
primus-zktls-js-sdk-dapp/ |
Node.js + Vite | Use the primus-zktls-js-sdk-dapp skill to scaffold a DApp that proves <what_to_prove>. The template ID is <template_id>. My appId is <app_id>. |
onchainos-verifiable-skills |
OKX OnchainOS API calls adapted into verifiable flows using primus-zktls-core-sdk. Includes DEX market/token, wallet portfolio, swap, and gateway endpoints with runnable demos. |
onchainos-verifiable-skills/ |
Node.js | Use OnchainOS Verifiable Skills to prove <endpoint_type> data from OKX OnchainOS with zkTLS attestation. |
binance-verifiable-skills |
Binance Skills Hub API calls adapted into verifiable flows using primus-zktls-core-sdk. Includes first-release Spot, Margin Trading, USDS Futures, and Alpha demos plus shared Binance signing helpers. |
binance-verifiable-skills/ |
Node.js | Use Binance Verifiable Skills to prove a Binance Spot, Margin, Futures, or Alpha API response with zkTLS attestation. |
primus-zktls-core-sdk |
Integration guide for Primus zkTLS Core SDK, including SDK calls, attestation generation, local verification, and Solidity on-chain verification with zktls-contracts. |
primus-zktls-core-sdk/ |
Node.js | Use Primus zkTLS Core SDK to explain how to build a request, generate an attestation, and verify it on-chain. |
web-template-capture |
Browser-driven template discovery for fields exposed through network responses or rendered DOM. | web-template-capture/ |
Node.js + Playwright | Use Web Template Capture to prepare a template for <site_url>. The target field is <field_name>. The known sample value is <sample_value>. |
Each skill directory is expected to contain its own SKILL.md. If the skill has runtime dependencies, prefer keeping a local package.json inside that skill folder.
- Skill artifacts should default to user-level data directories, not repository directories.
web-template-capturestores captured session data under~/.web-template-capture/artifacts/.- Skills that need persistent browser state should store it outside the repository.
- Do not commit generated artifacts, browser profiles, or session data.
- Keep the repository root lightweight and skill-agnostic.
- Put user-facing workflow details inside each skill directory.
- Prefer
setupanddoctorscripts for skills with external dependencies. - Keep root documentation focused on discovery, installation, and shared conventions.
When adding a new skill:
- Create a dedicated skill directory.
- Add a
SKILL.mdwith clear trigger and workflow instructions. - Add self-contained setup and diagnostics when the skill depends on external runtimes or tools.
- Update the Skills table in this README.