BankrBot/skills: a catalog of installable agent skills for crypto work
Bankr Skills equip builders with plug-and-play tools to build more powerful agents.
At a glance
- What is it?
- A JavaScript repository whose top level is a list of folders, each one a named skill an agent installs on its own, covering Uniswap, Base, Ethereum, token scanning and onchain research.
- Who is it for?
- BankrBot/skills is at its best as a reading list with an installer attached: the provider suites for Uniswap, Base and EthSkills are unusually specific about the traps in each protocol, and they give an agent concrete file paths to work from instead of a vague instruction to build a dApp. It is not a library, and nothing in the repository resolves as an npm dependency, so there is no version pinning between skills and no single upgrade path.
- Can I use it commercially?
- Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
- Is it still maintained?
- Yes. The repository last received commits 1 day ago.
- What is it written in?
- Mainly JavaScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on October 11, 2026, and from our analysis. They are not legal advice.
Editorial analysis
A folder per skill, installed by name
The repository description is short: Bankr Skills equip builders with plug-and-play tools to build more powerful agents. What that means in practice is visible in the install section, which is the entire installation procedure in a single line:
> install the [skill-name] skill from https://github.com/BankrBot/skills/tree/main/[skill-name]There is no package to add to a manifest and no dependency to resolve. The unit of distribution is a directory whose name is the skill's name, and you install exactly the ones you want. The tree is the catalog. Alongside the provider folders you will find names like `aeon-deep-research`, `aeon-vuln-scanner`, `bankr-token-scam-analysis` and `bankr-twitter-agent`, plus a `.claude/` directory at the top level and dozens of smaller community entries such as `0xwork`, `1claw`, `BOTCOIN`, `agent-wormhole` and `agenticbets`.
So the repository is better understood as a directory of capability folders than as a software project. The listed language is JavaScript, but that reflects the sample material inside some folders rather than a runtime that ships to you. When you install two of them, nothing checks that they agree with each other.
Three provider suites that carry most of the weight
Three suites are documented in enough detail to be genuinely useful, and between them they cover most of the ground a team building on Ethereum touches.
The Uniswap suite has five skills. `uniswap-cca` handles Continuous Clearing Auction contracts, including convex supply schedule generation, Q96 price calculations and multi-chain CREATE2 deployment. `uniswap-driver` plans swaps and liquidity positions, verifies tokens on chain, researches market conditions and produces pre-filled interface URLs across 12 chains. `uniswap-hooks` is the one to read first if you are writing v4 hooks, because it is framed around threat modeling, permission flag analysis, NoOp attack prevention and delta accounting. `uniswap-trading` covers V2, V3 and V4 execution through the Trading API, the Universal Router or direct contract calls, and `uniswap-viem` handles EVM plumbing through viem and wagmi, including multicall and event subscriptions.
The Base suite is five skills with a similar spread. `base-account` covers ERC-4337 smart wallets, one-tap USDC payments, gas sponsorship through paymasters and spend permissions across 9 chains. `base-deploy` is a Foundry deploy and verify path with testnet faucet access through CDP and BaseScan verification. `base-minikit` maps the old MiniKit OnchainKit approach onto the native Farcaster SDK, hook by hook. `base-network` is plain chain configuration, and `base-node` covers running a production node with Reth, including hardware sizing, snapshot bootstrapping and hardening.
The EthSkills suite is sixteen skills, which makes it the broadest. It runs from `ethskills-audit` and `ethskills-security` through `ethskills-gas`, `ethskills-indexing`, `ethskills-l2s`, `ethskills-qa` and `ethskills-orchestration`. Each sub-skill installs from its own folder, so you can take the audit checklist without the frontend playbook.
The aeon family is a different kind of tool
Set the provider suites aside and a second category shows up in the tree: the `aeon-*` folders, roughly twenty of them, and they are not protocol references. They are jobs an agent runs.
`aeon-deep-research`, `aeon-autoresearch`, `aeon-paper-pick` and `aeon-last30` are research and digest skills. `aeon-hacker-news-digest` and `aeon-rss-digest` are feed readers. `aeon-defi-monitor`, `aeon-on-chain-monitor` and `aeon-reg-monitor` watch positions or rules of thumb over time, and `aeon-monitor-kalshi`, `aeon-monitor-polymarket` and `aeon-monitor-runners` do the same for specific venues. Then there is a token layer: `aeon-token-pick`, `aeon-token-movers`, `aeon-distribute-tokens` and the aptly named `aeon-vuln-scanner`.
The three that stand out for anyone maintaining a skill catalog are `aeon-skill-evals`, `aeon-skill-repair` and `aeon-skill-security-scan`. Those three exist to evaluate skills, fix broken ones and scan them for problems, which tells you the authors assume the catalog needs exactly that maintenance. The `bankr-*` entries form a third group tied to the platform itself: `bankr/`, `bankr-communities`, `bankr-shopify`, `bankr-token-scam-analysis` and `bankr-twitter-agent`.
Read together, the tree has three layers with different failure modes. Provider skills rot when a protocol changes. Aeon skills rot when a data source moves. Platform skills rot when the platform does. Nothing in the repository marks which is which.
Payment rails run through the whole catalog
One design idea shows up in enough separate skills to be a deliberate choice rather than a coincidence: paying for data at call time instead of holding an account.
The `alchemy` skill describes EVM JSON-RPC, token balances, NFT ownership and metadata, transfer history, token prices, portfolio data, transaction simulation, webhooks and Solana RPC. It supports a plain API key, x402 pay-per-request authenticated with SIWE or SIWS and settled in USDC, and what it calls MPP on Tempo and Stripe. `zerion` reads interpreted wallet data across more than 41 chains, turning raw addresses into portfolio values, positions, PnL, NFT holdings, gas prices and swap quotes with USD values and protocol labels attached, and it prices a request at $0.01 USDC under x402.
`suwappu-dex` is the most cautious of the three. It routes across 14 chains and more than 10 providers including Li.Fi, CoW, Jupiter and Wormhole, exposes a native MCP server, and ships read-only by default with `execute_swap` gated behind an explicit step. That default is the interesting part, because an agent that can read balances but cannot trade is a much smaller blast radius than one that can do both.
`ethskills-orchestration` ties the same idea into a build flow with ERC-8004 and x402 for agent commerce. The pattern is consistent enough that you should treat any skill's payment requirements as part of its contract: read what it will spend, in what currency, and whether you can cap it, before you let an agent call it in a loop.
Repository shape, and what the numbers say
The metadata is worth reading closely because it describes a catalog rather than a library. There are 1,199 stars and 625 forks, which is a fork ratio near half, consistent with people copying a folder out and adapting it. There are 353 open issues. There are no releases, so there is no version to pin and no changelog to read. The declared license is null, which is an unusually important omission in a repository meant to be copied into other people's projects. The last push was on 2026-09-19, and the project is not archived.
That combination, actively pushed with 353 open issues and no releases, points to a repository where each folder is reviewed on its own schedule by whoever contributed it. The consequence for a reader is that quality varies between folders and there is no cross-folder consistency to lean on. When two skills disagree about how to read a balance, nothing in the repository resolves the disagreement for you.
Date sensitivity is the second thing to check, because several skills carry live state. `ethskills-addresses` publishes verified contract addresses for Uniswap, Aave, Safe, USDC, ENS, Chainlink, Lido and Aerodrome across mainnet and major L2s, and states that every entry was verified on chain through `eth_getCode` as of February 2026. That is a good practice and a dated artifact at the same time, and the age matters more for a contract address than for a threat modelling checklist. `ethskills-l2s` likewise folds in recent chain changes such as the Celo migration and the Polygon zkEVM shutdown. `ethskills-gas` is written around current gas economics. These are the folders to re-read when they matter most.
Where this is the wrong tool
The clearest limitation is that none of this is executable. If you expected an SDK with typed clients and versioned interfaces, this is not that, and the JavaScript language label can mislead you into expecting one. A skill folder is guidance: it tells an agent which files to read, which patterns hold on a given protocol and which mistakes recur. That is useful up to the point where your situation differs from the one the skill was written for, at which point you are on your own.
The second limitation is the payment model. Skills that can spend USDC, sign in with SIWE or SIWS, or hold a paymaster key for gas sponsorship introduce real failure modes into an automated loop. An agent that retries a paid call can spend money on a service that was already answering. An agent that holds a sponsored gas allowance can act at a cost you did not budget. The `execute_swap` gate in `suwappu-dex` shows the maintainers thought about this; not every skill necessarily did.
The third is the security surface. `ethskills-audit` describes a 500 item audit system spanning 19 domains that runs parallel specialist agents, synthesizes results and files GitHub issues, and it is explicit that it targets contracts you did not write. `aeon-skill-security-scan` and `ethskills-security` cover the other direction. That is a lot of security-adjacent text arriving as installable skills, and the reasonable response is to treat all of it as input to a review rather than as a review. A checklist produced by an agent is a starting point for a human reading the code, not a substitute for one.
Editorial conclusion
BankrBot/skills is at its best as a reading list with an installer attached: the provider suites for Uniswap, Base and EthSkills are unusually specific about the traps in each protocol, and they give an agent concrete file paths to work from instead of a vague instruction to build a dApp. It is not a library, and nothing in the repository resolves as an npm dependency, so there is no version pinning between skills and no single upgrade path. The sharpest thing to check before trusting any of it is date sensitivity, since ethskills-addresses carries a February 2026 onchain verification stamp and 353 open issues sit against a catalog that mixes protocol reference material with live data feeds. Start with the folder that matches the protocol you are actually shipping on, read it end to end, and treat the payment rails and the security skills as the two areas where the cost of being wrong is highest.
Frequently asked questions
What is BankrBot/skills and how do I install a skill from it?
It is a repository of skill folders, each named for a capability, that you install individually by asking your agent to install the skill from its folder URL, in the form https://github.com/BankrBot/skills/tree/main/[skill-name]. There is no package manifest step and no dependency to add, and each sub-skill in a provider suite installs on its own.
Which chains does the Zerion skill in this repository cover?
It covers 41 or more chains and returns interpreted wallet data rather than raw addresses: portfolio values, token and DeFi positions, transaction history, PnL tracking, NFT holdings, gas prices and swap quotes, with USD values and protocol labels attached. It supports x402 pay-per-request at $0.01 USDC as well as ordinary API key access.
What does the ethskills-audit skill actually do?
It is a 500 item EVM smart contract audit system spread across 19 domains. It runs parallel specialist agents, synthesizes what they find and files GitHub issues, and the README is explicit that it is separate from the security skill because it is meant for auditing contracts you did not write.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/bankrbot-skills)