SkillAgentSearch skills...

token-scam-analysis

Deep on-chain scam / rug / soft-rug analysis for EVM tokens (especially Clanker, Doppler, Bankr-style single-admin ERC-20s)

Install / Use

npx skills add BankrBot/skills --skill bankr-token-scam-analysis

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

86/100

Category

Security

Supported Platforms

Universal

Our assessment of token-scam-analysis

token-scam-analysis scores 86/100 on our quality scale, 754th of 1,108 Security skills we index.

Its SKILL.md is 22 KB long, well organised into 12 sections and no code examples: a thorough specification that gives an agent plenty to work with.

With 1,201 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
30/30
Structure
13/20
Description
15/15
Adoption
13/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so token-scam-analysis is actively maintained.
  • Our last check on 2026-11-03 found the source still online.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 88/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-10-06. Automated pattern scan on 2026-10-06. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

token-scam-analysis compared with similar skills

All 4 of these similar skills score higher than token-scam-analysis; compare them before choosing.

SkillScoreStarsUpdatedFormat
token-scam-analysis (this skill)by BankrBot861.2k54d agoSKILL.md
x402by browser-use100121.0k3d agoSKILL.md
algorithmic-artby anthropics100177.9k55d agoSKILL.md
pptxby anthropics100177.9k55d agoSKILL.md
designby nextlevelbuilder100149.0k6d agoSKILL.md

Frequently asked questions

How do I install token-scam-analysis?
Run npx skills add BankrBot/skills --skill token-scam-analysis. The install tabs above show the steps for each supported agent.
Which AI agents does token-scam-analysis work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is token-scam-analysis safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 88/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is token-scam-analysis still maintained?
The repository was last updated about 2 months ago, so token-scam-analysis is actively maintained.

name: token-scam-analysis description: Deep on-chain scam / rug / soft-rug analysis for EVM tokens (especially Clanker, Doppler, Bankr-style single-admin ERC-20s). Use when the user asks to "analyze this token for scam", "is this a rug", "should I trust this migration", "do a deep dive on holders and deployer", or provides one or more token addresses and wants a risk verdict backed by on-chain facts. Especially useful for migration narratives where a team claims they are "redeploying to fix tokenomics". emoji: 🔍 tags: [scam, rug, analysis, forensics, clanker, doppler, evm, security] visibility: public

Token Scam Analysis Skill

You are performing a forensic on-chain analysis to determine whether a token (or set of tokens) is a scam, rug-pull, or soft-rug. The output is a written report saved to the user's file storage plus a verdict in chat.

When to use this skill

  • User gives one or more token contract addresses and asks for a risk assessment.
  • User mentions a "migration" / "redeploy" / "new contract" narrative.
  • User names a Twitter/X handle of the project and wants to cross-reference claims vs on-chain reality.
  • Comparing old vs new contracts from the same team.

Core principle

Narrative is noise. On-chain state is signal. Every claim the team makes should be checked against what the contract and the deployer's wallet actually did. If the two conflict, the chain wins.

BUT: on-chain cleanliness ≠ not a scam. A team can deploy a perfectly clean LayerZero OFT or ERC-20, hand it to a multisig, and still run a textbook insider pump-and-dump via CEX coordination and concentrated supply. You MUST always run the off-chain intel pass (Step Final-1) before issuing a verdict, or you will under-call real manipulation cases.


Step 0 — Read the platform's deploy docs BEFORE judging tokenomics claims

Before making any claim about what a team "could not" or "should have" configured, read the deployment docs for the launch platform. Otherwise you will miss capabilities and bait on the team's narrative. Identify the platform by looking at the allData() context field (e.g., "interface":"clanker.world"), the deploy factory address, or the token ABI (admin/originalAdmin/allData/isVerified is Clanker-style).

Priority reads by platform:

  • Clanker v4 (allData() context = clanker.world, factory 0xe85a59c628f7d27878aceb4bf3b35733630083a9):
    • https://clanker.gitbook.io/clanker-documentation/general/token-deployments — overview: 100B ERC-20, extensions up to 90% of supply.
    • https://clanker.gitbook.io/clanker-documentation/authenticated/deploy-token-v4.0.0 — full deploy payload (vault, airdrop, fees, up to 7 reward recipients, pool config).
    • https://clanker.gitbook.io/clanker-documentation/references/core-contracts/v4 — ClankerVault / Airdrop extension internals.
    • What you MUST know before judging:
      • Tokenomics (allocations, lockups, vesting, multiple reward splits, custom paired token, initial market cap, static/dynamic fees) are set in the deploy payload. A team cannot say "we had to redeploy because we couldn't configure tokenomics" — Clanker v4 supports all of that in one transaction.
      • Min vault lockup 7 days, min airdrop lockup 1 day.
      • Max 90% of supply across all extensions (rest goes to LP).
      • Token bytecode being identical across deploys is expected (it's a factory template) — do not use bytecode-sameness as a red flag by itself. Configuration differences live in the deploy payload, not in the token runtime code.
  • Bankr / Doppler / Scheduled Multicurve: deployer beneficiary shares, no migration, locked pool, 95/5 fee split.
  • Zora / Virtuals / Pump.fun: platform-specific, look for official docs via search_tool.
  • LayerZero OFT multichain tokens (source has import "@layerzerolabs/oft-evm/contracts/OFT.sol" and peers(uint32), setPeer, send, setEnforcedOptions): standard multichain pattern. setPeer by owner is the main live admin power — if signers collude they can add a malicious peer chain and mint via _credit. Existing peers matching across chain explorers = legit bridge config. LZ V2 endpoint on Base: 0x1a44076050125825900e736c501f859c50fE728c.

Save this step to your report. The "Claim vs reality" section needs to quote a specific docs capability the team could have used and chose not to.


Tool map (roughly in order of use)

  1. token_search (identifier_type=address, chain=...) — baseline market data: price, mcap, volume, 24h change, holder count, security scan flag. Set include_chart=false, include_market_data_image=false to keep context lean.

  2. get_token_launch_info — if the token was deployed via Bankr/Doppler, you get the deployer wallet + twitter handle + tweet URL for free. Always try this first even if you think it's Clanker; the tool returns cleanly on non-match.

  3. get_contract_abi (chain=...) — confirm the address is a real contract and enumerate read/write functions. Flag dangerous functions: mint, crosschainMint, setOwner, updateAdmin, blacklist, setFee, pause, updateImage, updateMetadata. For OFTs, note setPeer / setEnforcedOptions — these are owner-gated but not exit-scam primitives by themselves.

  4. read_contract for on-chain state. For Clanker v4 tokens:

    • totalSupply() view returns (uint256) — expect 100_000_000_000 * 10^18.
    • allData() view returns (address originalAdmin, address admin, string image, string metadata, string context) — the single richest read. Tells you deployer admin, current admin (if different = admin handoff happened), whether metadata/socials/audits are populated, and which interface launched it (clanker.world, Farcaster, etc.).
    • isVerified() view returns (bool) — Clanker's own verification flag.
    • balanceOf(address) view returns (uint256) — for any wallet you want to check (admin, top holders, pool manager). For OFTs: owner(), peers(uint32) for each known eid (Ethereum=30101, BSC=30102, Arbitrum=30110, Base=30184, Polygon=30109, Optimism=30111), endpoint(), msgInspector(), preCrime().
  5. get_clanker_reward_ownership (tokenAddress, chain) — returns every reward recipient/admin slot for the token. This is the single best signal for "did the team actually allocate ecosystem/marketing/CEX rewards like they claim" — if it shows only one {admin, recipient} both equal to the deployer, the team did NOT configure multi-recipient tokenomics.

  6. Direct viem reads via execute_cli when you need to batch-check balances, detect contract vs EOA (top holders with 48-byte bytecode are AA smart-wallet sniper proxies, not whales), or compute concentration. Install viem@2.21.55 and run a small analyze.mjs.

    RPC configuration (custom env var is OPTIONAL):

    • First, call get_env_vars to check if the user has set a custom RPC var for the chain you're analyzing (BASE_RPC_URL, MAINNET_RPC_URL, ARBITRUM_RPC_URL, etc.). If present, use it — it's almost always higher-rate-limit.
    • If no custom var is set, fall back to viem's built-in public RPC by creating the client with just the chain (no custom transport URL). Example:
      import { createPublicClient, http } from 'viem';
      import { base } from 'viem/chains';
      const rpcUrl = process.env.BASE_RPC_URL; // may be undefined
      const client = createPublicClient({
        chain: base,
        transport: rpcUrl ? http(rpcUrl) : http(), // http() with no arg uses chain.rpcUrls.default
      });
      
      This works for base, mainnet, arbitrum, optimism, polygon, bnb, unichain, etc. out of the box.
    • The public RPCs are rate-limited and can be flaky under load. If you see 429 / timeouts when batching many reads, either (a) add small delays between calls, (b) chunk the reads, or (c) tell the user at the end of the report that a custom RPC env var would make re-runs faster and more reliable — don't block the analysis on it.

    Always use this viem step to:

    • Cross-check totalSupply, admin, originalAdmin, isVerified, allData from the RPC directly (don't trust only indexer outputs).
    • getBalance + getTransactionCount for each deployer and the claimed Twitter wallet.
    • getBytecode on each top non-pool holder — 48 bytes = EIP-7702 / minimal-proxy smart wallet (sniper pattern); >100 bytes = real contract (could be Clanker vault/airdrop extension, or a Gnosis Safe); no code = regular EOA.
    • For Safe multisigs: Safe contracts have a known bytecode prefix and a VERSION() / getOwners() function — you can detect them specifically.
    • For cross-chain tokens: do balance reads on the OTHER chains too (e.g., if it's on Base but deployed as an OFT, read the Ethereum and BSC mirrors via their RPCs — again, custom env var if set, else viem public default).
    • Concentration math: sum top-20, subtract pool, compute non-pool concentration.
  7. market_intelligence with query_type="holders" (chain, token_id=contract address, limit up to 20) — top holder distribution. The #1 holder is almost always the Uniswap v4 PoolManager (0x498581ff718922c3f8e6a244956af099b2652b2b on Base). Always identify pool addresses first and exclude them from concentration math.

  8. get_chain_activity_for_wallet on each interesting wallet:

    • The deployer/admin for each contract — look for the lifecycle: funded → deploy → extract → (sometimes) claim fees.
    • Top 3 non-pool holders — were they fresh wallets that sniped at genesis? How were they funded?
    • The claimed Twitter wallet — does it have any footprint on the deploy chain at all?
    • For OFT tokens: check if top holders are sending TO exchange deposit wallets (Coinbase, Binance hot wallets) — this is the distribution-into-bid signal.
  9. get_evm_address_from_twitter_username / get_evm_address_from_farcaster_username — resolve the claimed social handle to a wallet. Then check with get_chain_activity_for_wallet and viem getTransactionCount whether that wallet is the admin of any contract or has any footprint on the deploy chain. Most rugs have social wallets with ZERO activity on the chain they're scamming.

  10. browse_url on basescan/etherscan/polygonscan (https://basescan.org/address/{addr}) only as fallback when get_contract_abi fails or to confirm an address is a DEX/pool/infra contract. Many docs sites (gitbook-hosted root pages) return empty via firecrawl — go directly to the specific article URLs. After extracting the fact you need, immediately prune_messages(tool_names=["browse_url"], replacement="<one-line fact>").

  11. perform_technical_analysis (only if user is Bankr Club — will error otherwise, don't retry).


Step Final-1 (MANDATORY) — Off-chain intel pass before verdict

A contract can be on-chain-clean and still be a running manipulation case. Before writing the TL;DR verdict, you MUST run a parallel off-chain pass. Run these three calls IN PARALLEL in one turn:

  1. search_tool for investigator / press coverage:
    • Query 1: "<TOKEN_SYMBOL>" <token_name> scam rug investigation twitter
    • Query 2: ZachXBT <TOKEN_SYMBOL> manipulation (ZachXBT is the most-trusted on-chain investigator in crypto — always check for direct callouts by name)
    • Query 3: <TOKEN_SYMBOL> pump dump exchange investigation <current_year> (to catch Bitget/Binance/OKX/Gate investigation announcements)
    • Query 4 (if a team name surfaces): "<founder name>" <token_name> founder (to identify the team and any prior-project baggage)
  2. get_social_sentiment_for_ticker — pass the ticker and additionalContext describing the token, chain, and whether it's in a parabolic move. Returns community split, funding-rate context, and surfaced allegations.
  3. browse_url on the project's official X/Twitter account if known (format: https://x.com/<handle>) — look for self-warnings from the project itself. A team publicly "warning

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars1.2k
CategorySecurity
Updated1mo ago
Forks629

Languages

JavaScript

Trust signals

88/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 medium