openclaw + gank
openclaw is an ai agent that runs on your machine and talks to you through telegram, discord, whatsapp, whatever. it supports mcp servers as skills — which means you can plug ingank-mcp-server and your agent gets access to everything: launches, trading, wallets, volume bots, privacy swaps. fully autonomous.
this is the setup that lets an ai run your entire token operation end to end.
the idea
what you need
- openclaw installed (mac/windows/linux)
- a gank account (gank.dev)
- an api key from the gank dashboard
- some sol in a wallet
getting started
1. install openclaw
grab it from openclaw.ai. follow their setup — pick your llm provider (anthropic, openai, local), connect your chat apps. their docs cover this better than we can.2. get a gank api key
go to gank.dev → profile → api keys → generate. copy it. you’ll paste this into the mcp config.3. add gank as an mcp skill
openclaw supports mcp servers natively. addgank-mcp-server to your openclaw mcp config:
4. verify it works
open a chat with your agent and ask it something like:“list my gank wallets”if it comes back with your wallet list, you’re connected. if it doesn’t have any wallets yet:
“create a dev wallet and a regular wallet on gank”
5. fund and go
send some sol to your dev wallet. from there the agent can do everything:“launch a token called TESTCOIN with ticker TEST, dev buy 0.01 sol”
“buy 0.05 sol of [mint address] from my regular wallet”
“start a volume bot on [mint] with 5 volume wallets, medium intensity, 30 minutes”
“vamp everything back to my dev wallet”you get the picture.
skills to look for
openclaw has a skill/plugin ecosystem. some things that pair well with gank:- token scanner / new pair alerts — feed new launches into the agent so it can auto-snipe based on criteria you set
- price monitoring — the agent watches positions and sells at your targets
- sentiment analysis — scrape twitter/telegram for alpha, let the agent decide what to ape
- scheduling — run volume bots at specific times, launch at peak hours
- portfolio tracking — the agent keeps a running tally and rebalances
example strategies
things people are doing with this setup. not financial advice, just patterns.launch + pump + extract
- agent creates all wallet types (dev, bundle, snipe, swarm, volume)
- funds them from a main wallet
- launches token with dev buy + bundle buys
- snipe wallets hit next block
- volume bot runs for 30 min
- swarm buy for extra pressure
- monitors price, sells at target
- vamps everything back to dev
snipe + ride + exit
- you or a scanner skill spots a new token
- agent snipes it immediately
- monitors price with trailing stop
- sells at profit target or cuts losses at stop
privacy rotation
- agent vamps all wallets to dev
- initiates privacy swap (sol→xmr→sol) to fresh wallets
- waits ~45 min for swap completion
- continues operating from the new wallets
tips
- start small. give the agent 0.1 sol and let it trade. watch the live feed. get comfortable before scaling up.
- use the dashboard to monitor. everything the ai does shows up in real time.
- the agent’s wallets are isolated from yours. it can’t drain your human wallets.
- openclaw has persistent memory. tell it your preferences once (“always use 20% slippage on sells”, “never spend more than 0.5 sol per trade”) and it remembers.
- if something goes wrong, you can always use the dashboard to manually sell or vamp. human overrides always work.
- revoke the api key from the dashboard if you need to kill access instantly.
links
- openclaw — the agent
- gank-mcp-server on npm — the mcp server
- gank dashboard — monitor everything
- mcp tools reference — full tool list