Skip to main content

What are agent skills?

Agent skills are reusable instruction files that teach AI agents how to use a specific tool or API. When you install the DexPaprika skill, your AI agent gets a complete reference of every endpoint, CLI command, streaming parameter, and common workflow — so it can make correct API calls without guessing.
Skills work with Claude Code, Cursor, Cline, OpenCode, and many other AI agents. One install command. No API key needed.

What’s included

The DexPaprika skill installs the following files into your project:
FileWhat it provides
SKILL.mdFull API reference: all REST endpoints, CLI commands, MCP setup, streaming API, SDKs, common workflows, token addresses, pagination, error handling
references/openapi.ymlComplete OpenAPI 3.1 specification with all schemas, parameters, and response types
references/cli-reference.mdEvery dexpaprika-cli command with flags, examples, and output formats
references/streaming-api.mdSSE streaming documentation: GET/POST endpoints, response fields, connection handling
Your AI agent reads SKILL.md automatically. The reference files are loaded on demand when the agent needs deeper detail (like exact response schemas or CLI flag options).

Install

One command

npx skills add github.com/coinpaprika/skills/ --skill dexpaprika-api
This copies the skill files into .claude/skills/dexpaprika-api/ in your current project.
Terminal output showing npx skills add github.com/coinpaprika/skills/ --skill dexpaprika-api completing successfully, with files copied to .claude/skills/dexpaprika-api/
To install both CoinPaprika and DexPaprika skills at once, drop the --skill flag:
npx skills add github.com/coinpaprika/skills/

What happens

  1. The skills CLI (by Vercel Labs) fetches the skill from GitHub
  2. Files are copied to .claude/skills/dexpaprika-api/
  3. On next startup, your AI agent auto-discovers and loads the skill
Your project directory should now look like this:
.claude/skills/
  dexpaprika-api/
    SKILL.md
    references/
      openapi.yml
      cli-reference.md
      streaming-api.md
File tree showing .claude/skills/dexpaprika-api/ directory containing SKILL.md and references folder with openapi.yml, cli-reference.md, and streaming-api.md

Restart your agent

After installing, restart Claude Code (or your AI tool) so it picks up the new skill files.

Verify it works

After restarting, ask your AI agent a DexPaprika question:
What blockchain networks does DexPaprika support?
The agent should return a list of supported networks (Ethereum, Solana, Base, Arbitrum, Polygon, and more). If it does, the skill is loaded and working. Try a more specific query:
Get the current price of WETH on Ethereum using DexPaprika
The agent should call https://api.dexpaprika.com/networks/ethereum/tokens/0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2 and return the price.
Claude Code answering a DexPaprika query about blockchain networks after skill installation, showing live API data returned

Example queries

Once the skill is loaded, your agent knows how to use the full DexPaprika stack. Here are some things you can ask:

REST API queries

Show me the top 5 pools on Ethereum by 24h volume
Get 30-day OHLCV data for the USDC/ETH pool on Uniswap V3
Search for PEPE token across all chains

CLI commands

The skill teaches your agent to use dexpaprika-cli if it’s installed:
Use dexpaprika-cli to stream live prices for WETH on Ethereum
Run dexpaprika-cli to find newly created pools on Solana

Batch pricing

Get prices for WETH, USDC, and DAI on Ethereum in a single request

Streaming

Write a Python script that streams real-time WETH prices using DexPaprika's SSE endpoint

Skills vs MCP server vs Claude Code plugin

Agent SkillsHosted MCP ServerClaude Code Plugin
Installnpx skills addAdd URL to config/plugin install
What it gives the agentFull API knowledge (endpoints, CLI, streaming docs)Live tool calls (14 MCP tools)MCP tools + agents + skills
Agent makes API callsYes, directly via REST or CLINo, MCP server handles callsMCP server handles calls
Works offlineYes (skill files are local)No (needs mcp.dexpaprika.com)No (needs MCP server)
Works withClaude Code (default), Cursor, Cline, and more via --agent flagClaude Desktop, Cursor, VS CodeClaude Code only
Best forAny agent that can make HTTP calls or run CLI commandsQuick AI IDE setup with managed toolsClaude Code users who want everything bundled
You can combine approaches. Install the skill for API knowledge and connect the MCP server for tool-based access. They complement each other.

Manage skills

List installed skills

npx skills list

Remove the skill

npx skills remove dexpaprika-api

Update to latest version

npx skills update

Troubleshooting

Restart your AI agent (Claude Code, Cursor, etc.). Skills are discovered at startup. If it still doesn’t load, check that the files are in the right location:
ls .claude/skills/dexpaprika-api/
You should see SKILL.md and a references/ directory.
The skills CLI requires Node.js. Make sure you have Node.js 16+ installed:
node --version
If Node.js is installed and it still fails, try running with a fresh npx cache:
npx --yes skills add github.com/coinpaprika/skills/ --skill dexpaprika-api
Be specific in your prompt. Instead of “What’s the price of ETH?”, try “Using the DexPaprika API, get the price of WETH on Ethereum.” The agent should recognize the skill context and use the correct endpoints.
Skills must be direct children of .claude/skills/. If the files ended up nested differently (e.g., .claude/skills/skills/dexpaprika-api/), move the dexpaprika-api folder so it sits directly inside .claude/skills/. See the directory tree in the Install section for the expected layout.

Next steps

CLI guide

Install dexpaprika-cli for direct terminal access to all endpoints

Hosted MCP Server

Add MCP server access to Claude Desktop, Cursor, or VS Code

Claude Code Plugin

Get the full plugin bundle with MCP tools, agents, and skills

REST API Reference

Complete endpoint documentation with interactive playground

Get support

Join Discord

Connect with our community and get real-time support.

Give Feedback

Share your experience and help us improve.

FAQs

No. The DexPaprika API is public and free. No keys, no registration. The skill teaches your agent to call the free API directly.
Skills give your agent knowledge (API docs, CLI commands, common patterns). MCP gives your agent tools (callable functions that return live data). Skills make your agent smarter about the API; MCP does the calling for it.
Claude Code, Cursor, Cline, OpenCode, Aider, and many others. The skills CLI auto-detects your agent, or you can target one explicitly with the --agent flag. See the full list of supported agents.
Yes. Install the skill for deep API knowledge and connect the MCP server for tool-based access. They complement each other well.
No. The skill files are static reference documents stored locally in your project. Your agent reads them locally. API calls only happen when your agent decides to query the DexPaprika API based on what it learned from the skill.