Agent Skill
Teach any AI agent to launch and research Runitup tokens in one step.
Give your AI agent one file and it knows how to launch tokens on Runitup, look up prices, holders and trades, and check a wallet's portfolio — without you explaining any of it.
It covers the rules that are not guessable from the API's shape: that a tokenized-equity pair only works on Uniswap V4 and reverts on the V3 venues, that pairing against one needs an ERC-20 approval before the launch where an ETH pool does not, and that those equities exclude US persons. Each of those otherwise turns into a reverted transaction and a confused agent.
https://runitup.gg/skills/runitup/SKILL.mdIt's plain markdown with standard frontmatter, so it works in anything that loads skills: Claude Code, OpenCode, Hermes, OpenClaw, Cursor, or your own agent loop.
Your keys stay yours
The skill teaches an agent to build transactions, never to sign them. It explicitly instructs the agent never to ask you for a private key or seed phrase, and to confirm the details with you before broadcasting anything that costs money.
Install it
Claude Code
Per project:
mkdir -p .claude/skills/runitup
curl -o .claude/skills/runitup/SKILL.md \
https://runitup.gg/skills/runitup/SKILL.mdOr once, for every project on your machine:
mkdir -p ~/.claude/skills/runitup
curl -o ~/.claude/skills/runitup/SKILL.md \
https://runitup.gg/skills/runitup/SKILL.mdThen just ask: "launch a token called My Coin, ticker MYC, one billion supply".
OpenCode
mkdir -p .opencode/skills/runitup
curl -o .opencode/skills/runitup/SKILL.md \
https://runitup.gg/skills/runitup/SKILL.mdCursor
Save it as a project rule:
mkdir -p .cursor/rules
curl -o .cursor/rules/runitup.md \
https://runitup.gg/skills/runitup/SKILL.mdHermes, OpenClaw, and anything else
Most frameworks either take the URL directly or read markdown from a skills directory. If yours does neither, paste the file's contents into your system prompt — it is self-contained and about 11 KB.
No framework at all
Fetch it into context at the start of a conversation:
curl https://runitup.gg/skills/runitup/SKILL.mdLet an agent find it on its own
If your agent browses, point it at the domain and it will discover everything from llms.txt — the convention agents check when they land on a site:
https://runitup.gg/llms.txtThat file links the skill, the API reference, the live config endpoint, and the rest of these docs. Telling an agent "read runitup.gg/llms.txt" is usually enough.
What the agent can do with it
| Ask for | It uses |
|---|---|
| "Launch a token called X" | POST /api/v1/launch/prepare, then your wallet signs |
| "Launch X paired with NVDA" | Same, with quoteToken and venue: "uniswap-v4" |
| "What's trending on Runitup?" | GET /api/tokens — every chain, or one with ?chain= |
| "How is $MYC doing?" | GET /api/tokens/{address}?chain=… |
| "Chart the last day of $MYC" | GET /api/tokens/{address}/candles |
| "What has this wallet launched?" | GET /api/wallet/{address}/launches |
| "What's in my portfolio?" | GET /api/wallet/{address}/portfolio — all chains at once |
| "What chains can I launch on?" | GET /api/v1/config |
No API key, no sign-up, no rate limit to negotiate. See the API reference for the full surface.
Keeping it current
The skill is served from this site, so it changes when the API changes. Re-run the same curl to
update — there's nothing to reinstall and no version to pin.
Worth re-pulling if you installed it before 2026-08-30: the old copy named a testnet as the only chain, and knew nothing about pairing against a tokenized equity or the venue rule that goes with it. An agent working from it would prepare a launch on the wrong network.