Skip to main content
Mako supports 8 EVM chains for on-chain queries, but Base is the home context. The model has a curated knowledge base covering Base-native protocols, tokens, launches, and ecosystem dynamics that general-purpose models lack.

​
What Mako knows about Base

The knowledge base includes curated entries for 15+ Base ecosystem projects, covering:
  • Protocol mechanics — how Aerodrome, Virtuals, Farcaster, and other Base protocols work
  • Token metadata — contract addresses, supply mechanics, launch details
  • Ecosystem context — relationships between projects, key builders, recent launches
  • Base-specific terminology — protocol names, token tickers, and community concepts that general models hallucinate on
This knowledge is maintained as structured entries with keyword matching, so Mako can surface accurate project information without relying solely on parametric model memory.

​
Example queries

These are the kinds of questions where Mako outperforms generic models: 
"what is happening with Virtuals launches on Base?"
Mako will search its knowledge base for Virtuals Protocol context, then optionally call web_search for recent activity, and synthesize an answer grounded in both. 
"compare Aerodrome and Uniswap liquidity for USDC on Base"
Mako can chain knowledge_search for protocol context with get_token_info and web_search for current data. 
"what did Base announce on X this week?"
Mako will use web_search to find recent Base announcements and read_tweet to pull specific posts. 
"is this contract verified, how old is it, and what does the deployer hold?"
Mako chains is_contract → get_tx_count → get_eth_balance across the deployer address.

​
Multi-chain support

While Base is the product identity, every on-chain tool works across all 8 supported chains:
ChainAliases
Ethereumethereum, eth
Basebase
Arbitrumarbitrum, arb
Optimismoptimism, op
Polygonpolygon, matic
Avalancheavalanche, avax
BSCbsc, bnb
Gnosisgnosis
The default chain for on-chain tools is ethereum for compatibility with the broadest set of developer use cases. To query Base, pass chain: "base" explicitly.

​
Knowledge base architecture

The knowledge system uses keyword-triggered retrieval-augmented generation (RAG). When a user query matches keywords associated with a knowledge entry, the entry is injected into the model’s system prompt before inference. This means:
  • Answers about known projects are grounded in curated facts, not model hallucination
  • The knowledge base can be updated without retraining the model
  • New projects can be added by appending entries — no fine-tuning required
See Knowledge Tool for the full technical details.