Glossary

Definitions of key terms used across Balchemy Hub, Explorer, integrations, and trading docs.

0x

An EVM swap aggregator used by Balchemy for EVM trades. Balchemy public docs focus on Base/EVM trading flows. See also: EVM Trading.

Agent

A software actor that can call Balchemy through scoped access. In public docs, this means a 24/7 Web3 agent operated from Hub. See also: Hub Overview.

Agent Card

A machine-readable document that describes an agent's public capabilities and discovery metadata. Agent cards are used by discovery and integration flows. See also: ERC-8004 Agent Identity.

agentId

The stable identifier for a Hub agent in agent operation records. It is different from legacy bot IDs and different from the public MCP publicId used in endpoint URLs.

Ask Bot

A legacy tool name for a conversational MCP/tool path that sends a natural-language message through an agent context and returns a response. Available behavior depends on the caller's key, scope, and runtime context. See also: MCP Integration.

Automation

A runtime pattern where a Hub agent reacts to defined rules, schedules, signals, or conditions. New user-facing automation should be expressed through Hub rules and runtime state. See also: Rules and Runtime.

Base

An EVM-compatible Layer 2 blockchain. Balchemy supports Base/EVM trading flows where available in the app and execution path. See also: EVM Trading.

Bot

A legacy Studio-era record or compatibility bridge. New user-facing product work should use Hub agents, not Studio bot creation or bot cockpit flows. See also: Studio Retired.

Bot MCP Key

A retired Studio-era MCP credential shape. Active user-facing MCP/API access is Hub agent-scoped. See also: Agent Keys.

Channel Policy

A legacy Studio-era channel configuration. Active Hub product work should use Signal Source configuration for Telegram/Discord token-signal ingestion, source health, submitter trust, and attribution. See also: Signal Sources.

Circuit Breaker

A safety mechanism that can halt order execution when a provider, chain, or execution path is unhealthy. Circuit breakers protect users from continuing into known-bad execution conditions. See also: Order Lifecycle.

Claim

An ownership-sensitive Hub action used when an operator needs to take or prove control over an agent's operational state. Claim or similar sensitive flows can require step-up style verification. See also: Hub Overview.

Controlling Address

The wallet address associated with ownership or control for an agent or sensitive operation. The exact meaning depends on the onboarding and verification flow. See also: Agent Wallet Management.

Custodial Wallet

A Balchemy-managed trading wallet used by authorized product flows. Wallet and key operations are fund-sensitive and must be handled through the product UI/API flow, not raw database edits. See also: Agent Wallet Management.

DCA

Dollar-Cost Averaging. A trading strategy that splits a buy or sell into smaller orders over time or conditions. See also: Order Lifecycle.

Discovery

The process of finding public Balchemy agents, strategies, or metadata through Explorer or discovery-compatible surfaces. See also: Explorer.

ERC-8004

An agent identity and discovery-oriented integration model used by Balchemy for agent onboarding and public agent metadata. See also: ERC-8004 Agent Identity.

Explorer

Balchemy's public discovery surface. Use Explorer to browse or share public agent/strategy metadata. Explorer is not the main operating workspace; Hub operates 24/7 Web3 agents. See also: Explorer.

Fee-share

A configured attribution or partner-fee concept for Hub Signal Source or partner contexts. Fee-share is separate from referral rewards. Public docs should not promise automatic payout timing or settlement rails unless the active product flow exposes them. See also: Trading Fees.

Global Swap Fee

Balchemy's standard fee model on executed swaps. It is not a subscription, product tier, Stripe plan, or feature gate. See also: Trading Fees.

Guardrail

A safety constraint applied before or during trade execution, such as size limits, slippage limits, allowed token rules, policy checks, or risk controls. See also: Order Lifecycle.

Hub

Balchemy's operations surface for 24/7 Web3 agents. Use Hub when an agent needs identity, funding, scoped MCP/API access, logs, monitoring, key rotation, and ongoing operational control. Hub is not a separate pricing tier. See also: Hub Overview.

Hub Agent Key

A scoped MCP/API key for a Hub agent, managed from Hub → Agents & Keys and the agent detail flow. The old /hub/api-keys route redirects to /hub/agents, so Agents & Keys is the source of truth. See also: Agent Keys.

JWT

JSON Web Token. Balchemy uses signed tokens in several authentication and agent identity flows. Treat JWTs as credentials and do not paste them into public prompts, logs, issues, or screenshots. See also: Security.

Jupiter

A Solana swap aggregator used in Solana execution flows. See also: Solana Trading.

Knowledge Base

Legacy Studio bot knowledge content. Do not remove backend helper knowledge or Hub agent memory just because a file or service uses the word knowledge; those surfaces are governed separately. See also: Hub Overview.

MCP

Model Context Protocol. Balchemy uses MCP to expose scoped tools to compatible clients. Active public MCP/API access is Hub agent-scoped. It is a protocol surface over the same execution core, not a separate product tier. See also: MCP Integration.

MCP Key

A scoped credential used to call Balchemy through MCP. A Hub agent key belongs to one always-on Web3 agent. Keys are revealed once and should be stored in a secrets manager. See also: Agent Keys.

Nonce

A single-use value used during wallet or identity verification to prevent replay. A nonce should be signed only in the intended app or trusted authentication flow. See also: Security.

Order Lifecycle

The state path a trade/order follows from request through policy checks, execution, confirmation, failure, or completion. See also: Order Lifecycle.

Pending Order

An order that has been created but has not yet been submitted on-chain. It may need approval, additional checks, or a condition to be met before execution. See also: Order Lifecycle.

Pre-trade Report

A trade preview or analysis generated before execution. It can include expected output, slippage, fees, and relevant safety context. See also: Solana Trading.

Primary Channel

A legacy Studio-era channel concept. Active Hub docs should use Signal Source for Telegram/Discord source authorization, token-signal ingestion, source health, and attribution. See also: Signal Sources.

publicId

A public identifier used in MCP endpoint URLs:

https://api.balchemy.ai/mcp/<publicId>

Use the publicId shown in the Hub flow. Do not guess it from logs or unrelated URLs.

RAG

Retrieval-Augmented Generation. A technique where relevant context is retrieved and supplied before an assistant answers. Legacy Studio bot knowledge and Hub helper/memory context are separate product surfaces.

Rate Limit

A reliability control that limits request volume over a time window. Balchemy rate limits protect app, MCP, and AI reliability; they are not paid access tiers. See also: Rate Limits.

Referral

A referral-code and referral reward/commission system. Referrals are separate from Signal Source or partner attribution. See also: Trading Fees.

Risk Policy

A set of trading constraints used to decide whether an order is allowed to continue. Risk policy checks protect users before fund-sensitive actions execute. See also: Order Lifecycle.

Scope

The permission level carried by an MCP key, API key, or identity token.

Scopes are permissions, not pricing plans. See also: Scopes and Access Control.

SIWE

Sign-In with Ethereum. A wallet-authentication pattern where an EVM wallet signs a message containing a nonce and domain. See also: Security.

SIWS

Sign-In with Solana. A Solana wallet-authentication pattern where a Solana wallet signs a message containing a nonce and domain. See also: Security.

Slippage

The difference between quoted and actual execution price. Balchemy uses slippage limits to prevent trades from executing outside the caller's tolerance. See also: Order Lifecycle.

Smart Order

An order that uses configured logic such as triggers or conditions before execution. See also: Order Lifecycle.

Solana

A high-performance Layer 1 blockchain supported by Balchemy trading flows. Solana trades use Solana wallet addresses and Solana execution infrastructure. See also: Solana Trading.

Step-up Authentication

An additional verification step required for sensitive actions, especially management or ownership-sensitive flows. See also: Agent Keys.

Strategy

The trading logic, rules, and behavior configured for an agent or surfaced through public discovery. Strategies can be operated from Hub and discovered through Explorer when public. See also: Explorer.

Studio

Studio is retired as a separate user-facing bot workspace. Legacy routes and docs may remain reachable during migration, but new product setup starts from Hub agents. See also: Studio Retired.

Tool

A server-executed capability exposed through Balchemy's internal tool registry and, when allowed, through MCP or LLM function calling. The visible tools depend on runtime registry state, environment configuration, key scope, and principal context. Use tools/list as the source of truth. See also: Tool Catalog.

Trading Engine

Balchemy's execution service for order lifecycle, on-chain trading, fee calculation, and fund-sensitive trading controls. Public docs should focus on behavior and safety rather than unverified performance claims. See also: Order Lifecycle.

Retired Embedded Chat

The retired Studio-era embedded bot surface for project websites. Hub does not expose an embedded chat product path. See also: Studio Retired.