The CryptoDataAPI Agent Skill teaches your AI agent how to read the crypto market with our data — which endpoint answers which question, how the tiers work, and ready-made workflows for regime reads, funding/positioning checks, liquidation-risk scans and daily snapshots. It ships as a standard Agent Skill (SKILL.md + reference files) that any skills-compatible agent can load.
Skills use progressive disclosure: the agent reads a short front-matter summary first and only pulls the deeper reference files when a task actually needs them, so the skill costs almost nothing until it's used. The format works across Claude Code, Claude Desktop, Claude on the web, Cursor, OpenAI Codex, Gemini and other agents that follow the skills.sh convention.
The skill is half of the pairing. The Agent Skill = knowledge (when and how to use the data); the MCP server = live tools (the actual data). Install both for the best results: the agent knows what to ask, and has the tools to ask it.
The fastest path is the skills CLI, which installs the skill globally so every project sees it. This pulls the maintained skill from our public GitHub repo.
The -g flag installs globally (drop it to install into just the current project); -y skips the confirmation prompt.
npx skills add Crypto-Data-API/cryptodataapi-skills -g -y
For Claude on the web (claude.ai) and Claude Desktop, install the skill as a packaged capability. Download dist/cryptodataapi-skill.zip from the GitHub repo, then upload it under Settings → Capabilities (Claude Desktop) or Settings → Capabilities / Skills (claude.ai).
Once uploaded, Claude will consult the skill automatically whenever a crypto-data task comes up. Pair it with the MCP connector (see the Claude (web) page) so it can act on what it knows.
You can also clone the repo and copy the skill directory into your agent's skills folder (for Claude Code that's ~/.claude/skills/). Or grab just the single-file SKILL.md with the legacy curl one-liner — handy for a minimal setup or a CI image.
What's inside: SKILL.md (the entry point with front-matter + core guidance) plus references/endpoints.md, references/workflows.md and references/tiers-and-limits.md — the deeper material the agent loads on demand.
git clone https://github.com/Crypto-Data-API/cryptodataapi-skills
mkdir -p ~/.claude/skills/cryptodataapi \
&& curl -s https://cryptodataapi.com/skill \
-o ~/.claude/skills/cryptodataapi/SKILL.md
“Use the CryptoDataAPI skill to summarize today's market: regime, health score, and Fear & Greed.”
“Following the skill's funding workflow, check whether BTC positioning looks crowded.”
“Which coins are flashing the highest liquidation risk? Use the liquidation-risk workflow.”
“What does the CryptoDataAPI free tier cover, and what needs Pro?”
They're complementary. The skill is knowledge — a SKILL.md that tells the agent which of our endpoints answers which question, how the tiers gate data, and proven multi-step workflows. The MCP server is the live tools — the actual calls that return prices, funding, regimes and so on. The skill makes the agent use the tools well.
You get the most value from both, but each works alone. The skill on its own still guides an agent that calls our REST API directly (for example via an HTTP tool). The MCP server on its own still works — the agent just discovers tools by trial rather than being taught the workflows. Install both when you can.
No. The skill is free and open-source. Running the workflows it describes uses your API key, and the Free tier (no card) already covers most market-wide questions — see pricing for what Pro and Pro Plus add.
Re-run npx skills add Crypto-Data-API/cryptodataapi-skills -g -y to pull the latest, or git pull in a cloned copy. For the zip-upload path, download the newest dist/cryptodataapi-skill.zip and re-upload it. We update the skill when endpoints or tiers change.
Missing an agent, a config that changed, or a step that didn't work? Tell us and we'll fix it.