How to Test a New Telegram Bot Safely
A step by step way to try a Telegram trading bot with a small balance: verify the handle, do one round trip, check the real fee, then get the money back out.
Published 2026-11-14
A trial run is one buy, one sell and one withdrawal
A trial run is not a demo mode. Telegram trading bots do not offer one. It is a short, deliberate sequence with real money at a size you can afford to lose:
- one small buy of a token
- one sell of that same token
- one full withdrawal of the remaining balance back to your own wallet
All three happen from a separate wallet that holds only that float, nothing else.
What the docs cannot show you
A bot's fee documentation gives the advertised rate, and its feature list gives the tick boxes. Both are static. Neither shows what is actually deducted when an order executes, how the bot reports that deduction in the chat, how quickly a fill lands on a thin pair, or whether a withdrawal arrives without a manual nudge.
A round trip on a small balance puts those things in front of you in a few minutes, with very little at stake. The advertised number and the deducted number are not always the same figure, and the chat message is where you find out.
Where the trial sits
The trial starts once a bot looks interesting and ends with a decision: scale up slowly, or walk away and leave the wallet empty. Checks made before any money moves come earlier than this. Sorting out a withdrawal that is already stuck comes later. The trial is the step in between, while the amount at risk is still small enough that either outcome is survivable.
The one line to remember
The exercise answers four questions: is the handle the genuine one, what does the bot actually charge, how well does it fill, and can you get the money back out.
The limit of a trial
Trading memecoins can lose all of the money put in. A separate trial wallet caps how much is exposed while you learn how a bot behaves. It does not remove the risk. A clean trial shows how the bot performed on one small pair at one moment; it says nothing about the next trade.
Step 1: Confirm the handle against the bot's own site
Open the bot's own website and copy the Telegram handle from there. That is the only acceptable source for the username you will search in Telegram. Not a screenshot, not a forwarded message, not a reply under a post.
Compare character by character
Impersonators copy the name and change one detail: a doubled letter, an underscore moved by one place, a capital where the real handle has none, or a suffix such as _bot added or dropped. Read the handle left to right, character by character, against the one on the site. A handle that looks almost right is the wrong handle.
Each project has its own pattern, which helps you spot a fake once you have seen the real one:
| Bot | Handle listed on its own site |
|---|---|
| Banana Gun | @BananaGun_bot |
| Trojan | @solana_trojanbot |
| Bloom | @BloomSolana_bot |
Banana Gun uses a capitalised name with _bot on the end. Trojan spells out its chain in lowercase before the bot suffix. Bloom keeps the chain in the name with no separator. Once you know which pattern the project uses, a handle that breaks it stands out.
What to ignore
Handles pasted in group chats, forwarded messages, direct messages, comment replies and sponsored search results are the usual delivery route for impersonator bots. Those places are where a fake handle spreads fastest, because anyone can post one and nobody checks it. Treat a handle from any of them as unverified even if it matches what you remember seeing.
Cross-check the documentation
Open the bot's documentation from a link on the same official site, and confirm it points to the same handle. Two pages on the project's own domain agreeing with each other is a stronger signal than one page alone. Do this before you send anything, including a deposit address.
The rule for this step
If anything differs, stop. Do not ask the bot itself to confirm it is genuine: a fake will happily tell you it is the real one.
Step 2: Build a wallet you can afford to lose
The wallet is the boundary of the test. Everything you are willing to risk sits in it, and everything you are not willing to risk stays out of it.
Create a new wallet for this trial only
Generate a fresh seed phrase for this purpose. Write it down on paper, offline, and store it away from the screen. Not a photo, not a password manager note, not a message to yourself in Telegram.
New seed phrase matters. A wallet reused from somewhere else carries approvals, history and any exposure it already has.
Keep it empty of everything else
No main holdings, no NFTs, no connections to other apps, no past transactions. A clean wallet is easy to read: when you check the token balance before and after the trade, every change is the bot's doing.
Fund it once, with a small amount
Size the balance to cover the token you are buying, the bot fee twice (you buy and you sell), and network costs on that chain. Pick the figure before you start and do not top it up mid-test. A top-up hides how much the round trip actually cost.
Match the wallet type to the chain
| Chain you are testing | Wallet type | Bots that fit |
|---|---|---|
| Solana | Solana wallet | Trojan, BONKbot |
| Ethereum, Base, BNB Chain | EVM wallet | Banana Gun, Bloom |
| Several chains supported | One chain only for the trial | GMGN, Maestro |
If a bot covers multiple chains, keep the test on one of them so fees and confirmations stay comparable.
Why the throwaway wallet is the point
If the bot turns out to be the wrong one, or malicious, the loss stops at that balance. Your main funds are not in the session at all. The same wallet also makes both connection routes in the next step safer, whether the bot asks for a private key or a wallet connection.
Step 3: Connect without handing over your seed phrase
Never type a seed phrase into a Telegram chat. Not for a bot, not for an admin, not for a "migration", not for a "verification step", not because a message says the bot needs it to restore your wallet. No legitimate setup needs your seed phrase. Anyone who has it owns every wallet that phrase controls, and can empty them without asking you again.
Importing a key and connecting a wallet are not the same
Both routes get a bot trading for you, but they hand over different things.
| Import private key | Connect wallet | |
|---|---|---|
| What the bot receives | The private key of that wallet | A connection and, later, token approvals |
| What the bot can do | Move everything in that wallet, at any time | Act within what you approved |
| What stays with you | Nothing, until you move funds out | Keys stay in your wallet app |
| How you undo it | Move the funds to a new wallet | Disconnect and revoke approvals |
Connecting keeps the keys in your wallet app, but the connection and any approvals you sign stay in place until you remove them. Neither route is "safe" in the abstract. What makes them survivable is Step 2.
Why the throwaway wallet does the work
With a key import, the bot holds a key to a wallet that holds almost nothing, so a bad outcome costs the trial balance and not your holdings. With a wallet connection, the approvals you sign are attached to a wallet you were going to empty at the end anyway. Your main wallet never appears in either flow.
Check what the bot asks for before you click
Read the bot's own documentation on setup first, then compare it with what the bot is asking you for. Bot pages such as /bot/sigma, /bot/bloom and /bot/banana-gun set out the chains and features each one supports, which is a useful baseline for judging whether a message is describing the real product.
Treat the chat as untrusted
Only act on instructions that match the bot's own website. Any link inside a chat asking you to "sync" or "validate" a wallet is a stop signal: disconnect, close the chat and start again from the site.
A bot that holds your private key can lose your funds. That is the whole reason this step exists.
Step 4: Place one small round trip
A round trip means one buy and one sell of the same token, in the smallest size the bot will accept. Before you touch anything, write down two numbers: your native balance (SOL or ETH) and the token balance, which should be zero in a fresh wallet. Note them again after the buy and after the sell. Those notes are what you will check the fee against in the next step.
On Solana
Open a Solana bot such as BONKbot, Trojan or Bloom, paste the token contract address into the chat, pick a small amount and confirm. The bot replies with a transaction link, then a confirmation showing the token amount received, the SOL spent and the network cost. BONKbot's fee docs state a 1% trade fee on Solana, Trojan's docs state 1%, and Bloom's fee docs state 1% on every supported chain, so the fee line should be roughly one hundredth of your buy. Sell the same token back through the same bot and note the SOL that returns.
On EVM
Open Banana Gun on Ethereum or Sigma on Base, paste the token address, set an amount and confirm. Banana Gun's own site states a 0.5% Ethereum trade fee; Sigma's fee docs state 1% on Base. The confirmation shows the ETH spent, the token amount received, and a gas line that sits apart from the bot's own cut. On a small trade, gas is often the larger of the two.
What you will notice
Solana confirmations are short, and the network cost is a minor line. On EVM the gas line is separate, varies with network conditions, and makes the message read longer. Both formats show a token quantity: check it against what you expected at that price.
Read the messages closely
- Did the bot state what you actually paid, or only the amount you requested?
- Did each confirmation arrive exactly once?
- Did any command fire twice, or produce two identical transactions?
If the fill fails
A buy that reverts is usually a slippage setting, not a broken bot. Change slippage deliberately, by a small step, and retry once. If the second attempt fails, stop the test and count it as a finding rather than adjusting further.
One round trip is the whole test. Do not add a second token or a larger size to get a better read.
Step 5: Check the fee the bot actually took
Do the arithmetic on your own trade rather than on the advertised rate. Start with the amount you put in, in the asset you spent. Take the amount that came back when you sold the same token. Subtract the price movement between entry and exit, and subtract the network costs you paid: gas, priority fee or tip. What remains is roughly what the bot took. Compare that remainder with the fee the bot's own docs publish.
A round trip pays the fee twice, once on the buy and once on the sell. Keep that in mind before you call the number wrong. The bot's cut is also separate from gas and from the token's price. A token that fell while you held it is not evidence of a fee problem.
What the published fees say
| Bot | Published trade fee | Source |
|---|---|---|
| Bloom | 1% on every supported chain, 0.9% with a referral | https://docs.bloombot.app/grow-with-bloom/fees-structure |
| BONKbot | 1% on Solana | https://docs.bonkbot.io/fee-structure |
| Trojan | 1% on Solana | https://docs.trojan.com/faqs |
| Sigma | 0% on Solana; 1% on Ethereum, Base, BNB Chain, Avalanche and Unichain | https://docs.sigma.win/other-info/fees |
Sigma's stated zero on Solana is useful for a first trial: it isolates the mechanics of buying, selling and withdrawing from the cost of doing so. Run the round trip there, get comfortable with the confirmation messages, then repeat it on a chain where the stated fee is 1% and check that the deduction matches.
When there is no published fee
BullX, Looter, MevX and Shuriken list no fee. Nothing to compare against means the live trade is the only place their cost becomes visible, which is exactly what the trial is for. Write down what went in and what came out, in the same asset, and keep the record next to your notes for the other steps.
If the remainder is far from what the docs state, repeat the round trip once with a slightly larger amount before drawing a conclusion. Very small trades are dominated by price movement and rounding, and the fee line hides inside that noise.
Before the first trade, put the bot's published fee next to its closest alternative on a side by side comparison, such as Banana Gun vs Trojan.
Step 6: Test the exit before you need it
A trial run is finished only when the money is back in a wallet you control. Withdraw whatever is left in the trial wallet, confirm it arrives, and check the balance on the receiving side rather than trusting the bot's success message alone.
Read the withdrawal screen before you confirm
The bot will show a minimum withdrawal amount and a network fee before you approve the transfer. Read both on screen instead of assuming they match what you expected. Leave enough in the wallet to cover the network fee, or the withdrawal will fail and you will have learned nothing except that the error message exists.
Close the door behind you
How you connected decides what cleanup looks like:
- Wallet connection: open your wallet's approval settings, review the permissions the bot created during the trial, and revoke them once the withdrawal has landed. A leftover approval can let a contract move tokens later without another signature.
- Private key import: assume the key is now known to the bot. Treat that wallet as spent, never reuse it for real funds, and move anything left to a fresh wallet.
Record the timings while they are fresh
Note how long a trade confirmation took, how long the withdrawal took, and whether anything needed a second attempt. These numbers are what you will compare against if the bot slows down later under a larger balance.
What the result means
A withdrawal that arrives cleanly, at the amount you expected, is the green light to consider funding the bot with more. A withdrawal that stalls, loops back to the start, or asks for extra steps you were not told about ends the test there. Do not add funds to find out whether the second attempt works.
Once you have tested on one chain, you can see what else runs on it: Solana bots and Ethereum bots.
What a trial tells you, and what it does not
What it tells you
One completed round trip answers a short list of concrete questions:
- Handle: the bot you traded with is the one the project's own site lists, not a lookalike.
- Fee: the amount actually deducted matches the figure published in the bot's own documentation.
- Messages: confirmations arrive, they are readable, and they state what happened rather than just "success".
- Fill: a small order goes through at a price close to what the confirmation shows, with no surprise extras.
- Exit: the remaining balance leaves the bot and lands back in your wallet without friction.
Those five answers are worth more than any amount of reading, because they are about behaviour rather than claims.
What it does not tell you
A test on a quiet afternoon says nothing about how the bot performs under heavy load or while a token is moving fast and messages queue up. One small trade says nothing about slippage or execution at larger size. A published fee can change after your test; nothing freezes it. And a trial shows nothing about the backend: who can reach the keys, how orders are routed, or what happens to your funds if the service goes offline.
Judge the process, not the result
A trial says nothing about whether trading will make money, and nobody can know that in advance. A small round trip that happens to end in profit is not evidence of anything.
| Round trip | Fees and messages | Withdrawal | Verdict |
|---|---|---|---|
| Small loss | Deduction matches the published fee, confirmations clear | Arrives promptly | Pass |
| Small profit | Confusing messages, unclear deduction | Slow or silent | Fail |
Then scale in steps
Move up in small increments only after every check passes. Keep your main wallet out of the bot until it has earned a place in your routine.
Red flags that end the test
A trial run is meant to surface problems while the balance is still small. Some findings are worth noting and continuing. The ones below are not: they end the test.
The bot asks for your seed phrase
No legitimate bot needs it, whatever the framing: verification, migration, a "security fix", a wallet restore. The same applies to a request for a private key sent through chat. Stop there.
The handle does not match the website
If the username differs by even one character from the handle on the bot's own site, you are not talking to the bot you researched. Treat a support account that messages you first, offering to help, as part of the same problem. Real support does not open the conversation.
Money moves without your command
A balance that changes on its own, or a confirmation for a trade you never placed, means someone else is operating the wallet. Withdraw what you can and stop.
The exit does not work
A withdrawal that stalls, is refused without explanation, or comes with a request to deposit more first is the most expensive warning you will get. Take it at face value.
A link to "sync", "validate" or "unlock" the wallet
These are wallet-draining patterns, not bot features. No trading bot needs you to unlock a wallet through a link.
Fees that do not match the published figure
If the deduction is higher than the fee in the bot's own docs and nothing there explains the difference, note the gap. For bots that publish no fee, such as BullX, Looter, MevX and Shuriken, there is no figure to compare against, so an unexpected deduction has no cover at all.
What to do
On any of these: stop the test, withdraw whatever still moves, abandon the trial wallet, and never reuse it for your main funds.
The trial checklist: before, during and after
The protocol above fits on one page. Use it as the written record of the trial: each row is a pass or a fail, not an impression.
| Stage | What to check | Why it matters |
|---|---|---|
| Before the first trade | Handle copied from the bot's own site and compared character by character; fresh wallet holding only a small balance; the bot's published fee read in its own docs | Impersonator bots copy names and logos, and the published fee is the number you will hold the live deduction against |
| During the round trip | One buy and one sell of the same token; the confirmation messages; token balance before and after | A second trade adds cost without adding information |
| After the withdrawal | Funds back in your own wallet; token approvals checked and revoked | An exit that works while the balance is small is the only evidence you have that it works |
What each row means
Before the first trade. Copy the handle from the bot's own site and compare it character by character with the one you plan to open. Banana Gun lists @BananaGun_bot, Trojan lists @solana_trojanbot. Fund a wallet created for this trial only, and read the fee in the bot's docs so you have a number to check the live deduction against.
During the round trip. One buy and one sell of the same token is enough. Note the balance before, read the confirmation messages, note the balance after.
After the withdrawal. Withdraw the rest, confirm it arrives in your own wallet, then revoke approvals.
The closing rule
Every row passing means scale up slowly, and only after the first small round trip has settled. Any row failing means walk away. No single good result cancels a failed exit test.
Once the trial is finished, put the bot you tested next to the one you nearly chose on a comparison page such as Bloom vs Maestro before you move in more funds.
Frequently asked questions
Can I test a Telegram trading bot without risking any money?
Partly. You can check the official handle, read the bot's own documentation, and watch how the interface behaves before you fund anything. But every trading bot in this category executes on-chain trades, and on-chain trades cost gas or network fees even when the swap is tiny. Some bots state their trading fee openly, for example Bloom's fee docs list 1% per trade (0.9% with referral) and BONKbot's fee structure states 1% on Solana, so a test trade will be charged that fee on top of network costs. Treat the first trade as a small paid test, not a free trial, and only put in money you can lose.
How do I know the bot handle is the official one?
Get the handle from the bot's own website or documentation, never from a Telegram search result, a forwarded message or a comment. Compare character by character: impersonator bots copy names and avatars and swap one letter or an underscore. A sourced example: Banana Gun's site points to @BananaGun_bot, BONKbot's docs point to @bonkbot_bot, and Trojan's docs point to @solana_trojanbot. If the handle you were sent differs in any character, stop. Also check that the bot asks for a wallet connection or deposit address in the way its docs describe, and never enter a seed phrase into any bot for any reason.
Do I need a separate wallet just for testing?
Yes, use a dedicated wallet. A Telegram bot that holds your private key can lose your funds, so the wallet you connect to a new bot should hold only the amount you are prepared to test with, never your main holdings. Create a fresh wallet, fund it with a small balance, and keep your long-term wallet unconnected. That limits damage from a bug, a wrong contract address or a compromised bot. It also makes the trial easier to read: the balance in that wallet is your full test budget, so you can see exactly what the bot deducts.
How small should the first trade be?
Small enough that losing the whole amount changes nothing for you. There is no universal figure, and it depends on the chain's minimum trade size and network fee, which vary. Start with a single trade at the smallest size the bot accepts, confirm the token arrives in your wallet, then check the fee the bot actually took against what its docs state. Only increase size after the full cycle works: buy, hold, sell, withdraw. Do not add funds mid-test to "make it worthwhile". One clean small round trip tells you more than a large one.
Should I revoke approvals after the trial?
Yes. Token approvals let a contract spend from your wallet, and a leftover approval stays live after you stop using the bot. After the trial, revoke approvals you no longer need using a reputable approval-revocation tool for that chain, then move any remaining balance out of the test wallet. Revoking costs a network fee, so budget for it. If you plan to keep using the bot, keep only the approvals needed for that bot and review them periodically. This is a routine hygiene step, not a sign the bot did anything wrong.
What if the bot's deducted fee does not match its docs?
Work it out before you accuse anyone. A deducted amount can include the bot's stated fee plus network fees, priority fees, slippage on the swap and any token transfer tax, so compare the actual charged fee line, not the total balance change. If the numbers still do not match the docs, stop trading, withdraw the remaining balance, and re-read the current docs; fee pages change. Then decide whether to continue. Bots whose fees are not published, such as BullX, Looter, MevX and Shuriken, cannot be checked this way, so observe the deduction on a small trade before scaling.