FAQ
General
What is Balchemy?
Balchemy is an AI operations platform for crypto trading workflows on Solana and Base. The product has two primary operating surfaces:
- Hub for owners/operators of 24/7 Web3 agents that need identity, funding, scoped MCP/API access, logs, monitoring, rules, Signal Sources, and Revenue attribution.
- Explorer for browsing verified public agent and strategy metadata.
Studio is retired as a separate bot workspace. Legacy Studio routes or docs may remain reachable during migration, but they are not the active product path.
What happened to Studio?
Studio is retired as a separate user-facing product. New setup starts from Hub agents, not Studio bot creation.
The useful Studio-era ideas are being transformed into Hub-native concepts:
- natural-language rules and runtime state,
- Signal Sources for Telegram/Discord token-signal ingestion,
- activity, positions, trades, and Revenue attribution,
- Hub agent-scoped MCP/API access.
Retired Studio surfaces include Studio bot creation, Studio bot cockpit, Studio knowledge/training, embedded chat, and Studio bot-scoped MCP.
Is there a paid plan?
No. Balchemy does not use product plans, feature tiers, Stripe billing, or locked product access. Every user gets the same standard product access.
Balchemy earns from executed swaps through the global swap-fee model. If you do not trade, there is no monthly platform charge.
Is Explorer a separate workspace?
No. Explorer is for discovery. Use it to browse public agents, strategies, and metadata. Use Hub to operate 24/7 Web3 agents.
Do Hub users paste Telegram or Discord bot tokens?
No. Hub Signal Sources use Balchemy-managed platform integrations and account/source verification flows. Do not paste raw Telegram or Discord bot credentials into the product UI.
Does Hub create a Studio bot for every agent?
No. Any linked legacy bot record is compatibility state, not a user-facing product promise. Hub agent lifecycle, runtime, rules, access, and Signal Sources are the active product surfaces.
What data does Balchemy collect?
Balchemy stores the data required to operate the product, such as:
- Wallet addresses used for authentication, ownership, funding, or trading context.
- Conversation, agent activity, and operational logs needed for continuity and troubleshooting.
- Trade/order records needed for execution, reconciliation, and analytics.
- Key, scope, and audit metadata for MCP/API access.
Do not paste secrets, seed phrases, private keys, or production credentials into chats, docs, issue trackers, or screenshots.
How do I delete my account?
Use the account deletion flow in the app when available, or contact support if the account has active trading wallets, pending orders, keys, or operational state that must be handled first. Do not try to delete or repair wallet/key state through raw database edits.
Signal Sources
What are Signal Sources?
Signal Sources are Hub-managed Telegram or Discord sources that can contribute token or contract-address signals, source health, submitter trust rules, and attribution.
Signal Sources are not chat-bot personality surfaces. Hub does not add generic Telegram/Discord auto-response behavior, knowledge Q&A, welcome messages, or mention-only community chat behavior.
Are Signal Source attribution and referrals the same thing?
No. They are separate systems.
- Signal Source attribution is tied to source-aware activity, Revenue attribution, or partner metadata where the active product flow exposes it.
- Referral rewards are tied to referral codes and referral reward/commission records.
A source can be tracked without that automatically meaning a live payout rail exists. Payout or settlement details depend on the configured product flow and should be verified before being promised.
Can a website embed Balchemy chat?
No active Hub product path exposes embedded bot chat. The Studio-era embedded chat surface is retired.
Where do I manage MCP keys?
Use Hub → Agents & Keys and the agent detail flow. Studio bot-scoped MCP is retired as a user-facing product path.
Hub
What is Hub for?
Hub is for operating 24/7 Web3 agents. It helps an agent owner/operator manage registration, setup state, funding, scoped MCP/API keys, logs, monitoring, and sensitive access changes.
Where are Hub agent keys managed?
Use 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 for Hub key management.
What scopes can a Hub agent use?
Hub and MCP use the same scope hierarchy:
| Scope | Use it for |
|---|---|
read | Status, logs, portfolio, research, and read-only calls. |
trade | Trade actions allowed by policy and risk checks. |
manage | Sensitive management actions such as key rotation or configuration changes. |
Use the smallest scope that works. Keep manage out of unattended runtimes unless it is absolutely required.
Does a Hub agent need human approval before every action?
Not every action. A properly scoped agent can make calls allowed by its key, setup state, and policy. Trading still passes through policy and risk checks. Sensitive management actions can require owner/claim or step-up style verification.
How does a Hub agent get funded?
Use the authorized Hub wallet/funding flow for the agent. Fund only the amount intended for active trading, keep staging and production funding separate, and never repair funding or wallet state with raw database edits.
Trading and fees
What are the trading fees?
Balchemy charges a standard swap fee on executed swaps. There are no monthly platform charges, product plans, or paid feature gates.
See Trading Fees for the public fee model and the distinction between global swap fees, Signal Source attribution, and referral rewards.
Which chains are supported?
Balchemy public docs focus on Solana and Base/EVM trading workflows. Solana trades route through Solana execution providers such as Jupiter. EVM trades focus on Base/EVM flows. Check the current trading docs and app UI for the exact chain list available to your account and agent.
What is the difference between a pending order and a submitted trade?
A pending order has been created but has not yet been submitted on-chain. A submitted trade has entered the on-chain execution path and is awaiting confirmation or failure. The History and order lifecycle views show the current status.
What happens if my trade fails?
If a trade fails before submission, it should not execute. If a submitted transaction fails on-chain, network costs or gas behavior depends on the chain and transaction state. Check the order details for the exact failure reason, such as insufficient balance, slippage, provider error, or policy rejection.
Integrations
How do Telegram and Discord verification differ from Signal Sources?
Hub verification links an operator identity for ownership, claim, authorization, or notification flows.
Signal Sources authorize a Telegram or Discord source for token-signal ingestion, source health, submitter rules, and attribution.
What happens when Discord OAuth fails?
The most common cause is a redirect URI mismatch between the Balchemy callback URL and the URL registered in the Discord application settings. Confirm the callback URL, protocol, hostname, and trailing slash behavior. See Discord Integration and Troubleshooting.
Is X/Twitter available?
Treat X/Twitter as not an active Hub Signal Source unless a future explicit product flow enables it. Do not rely on X/Twitter as a trading or execution channel.
MCP
Is MCP only for Hub agents?
Yes for the active public product. Hub agent MCP/API access is managed from Hub. Legacy Studio bot-scoped MCP may remain as compatibility code during migration, but it is retired as a user-facing product path.
What is the MCP endpoint?
https://api.balchemy.ai/mcp/<publicId>Use the publicId shown in the relevant Hub flow. Do not guess IDs from URLs or logs.
How many tools are available via MCP?
Do not rely on a hardcoded tool count. The visible tools depend on runtime registry state, environment configuration, key scope, and principal context.
Use MCP tools/list with the current key to see the exact tool catalog available to that caller.
What is the difference between agentId and publicId?
agentIdidentifies a Hub agent in operational records and agent-specific flows.publicIdis the public identifier used in MCP endpoint URLs and discovery surfaces.
Use the IDs shown by the app or onboarding flow instead of deriving them manually.