The NFT Floor Price Tracker Bot I Built on Telegram for Indian Web3 Founders

Here is my honest confession: web3 community questions pile up because founders and moderators are busy building, yet the same people keep answering the same floor-price and volume questions by hand. An NFT floor price tracker bot takes that repetitive checking off their plate without pretending it can predict the market.

It answers Telegram questions about floor price, sales, volume, listings and NFT data by calling OpenSea through the workflow platform. Its limitation matters just as much: it reports current information, but it cannot tell a holder what to buy or what will happen next.

Example:

The collection name and figures below are placeholders for illustration, not real market data.

You:  what's the floor on example-collection today, and is volume up?
Bot:  Floor price for example-collection is X ETH.
      7-day volume is Y ETH, higher than the previous 7 days.
      Want the cheapest 5 listings as well?

The holder asks in ordinary language. The workflow identifies the request, chooses the right OpenSea tool, retrieves live data and sends a readable answer back to the same Telegram chat.

For a founder or moderator, the relief is simple. They do not need to pause their work, open another tab, check a number and paste it into the group again.

What should an NFT floor price tracker bot never say?

The bot should never recommend buying, selling or holding an NFT.

If a holder asks, “Should I buy now?”, the assistant should respond with the available market data and explain that it cannot provide investment advice.

I write that boundary into the agent instructions and test it with pushy questions before handover. The rule protects both the holder and the collection team.

A message inside an official Telegram group carries the project’s name. A prediction such as “the floor will pump” can be screenshotted and shared as the collection’s position, even when nobody on the team approved it.

The bot answers “what is”. It never answers “what will be”.

Where does NFT market analysis stop?

NFT market analysis in this workflow means reporting information available through connected tools:

• Current floor price.
• Trading volume.
• Sales activity.
• Listings.
• Offers.
• Orders.
• Collection information.
• NFT metadata and traits.
• NFTs held by an account.

The workflow can compare the current figures returned by OpenSea. It can organise the answer and offer a related follow-up question.

It cannot promise that a price will rise. It cannot declare that a collection is undervalued, and it should not present a market opinion as a fact.

Founders sometimes feel pressure to keep the community excited. Moderators feel it too, especially when group members repeatedly ask where the market is going. The safer answer is useful data with a clear boundary rather than a confident prediction that may hurt trust later.

Why do I offer some workflow builds free?

Some of the people with the best ideas are not ready to pay for a custom automation.

I built this bot on one platform to show it working. A client can ask for any automation platform, or a custom build, and my team will set it up that way.

A student may be learning n8n while trying to build a useful project. Another person may be starting with zero or very little money and looking for a way to support their family. Somebody else may want automation for a community project that helps other people rather than creating quick personal profit.

I keep a free services option for eligible people in those situations.

I cannot accept every request. I look for passion, effort and a real willingness to learn how the workflow works after it is handed over.

People who only want to earn fast are not a fit. A free automation cannot create commitment where none exists, and the work has little value if the person disappears as soon as the build is complete.

Hard work matters more than polished words. A student who has tested a manual process, spoken with users or documented the repeated task shows more readiness than somebody asking for an instant AI business.

The free service exists because starting from zero is difficult. A useful workflow at the right time can give somebody the space to learn, serve a community or contribute to their family.

How does one Telegram message travel through the bot?

The design uses four workflows.

One main workflow receives Telegram messages and sends replies. Three sub-workflows act as specialised tools for analytics, NFT information and marketplace activity.

Workflow	Responsibility
Main workflow	Receives the Telegram message, selects a tool and sends the reply
Analytics tool	Retrieves collection statistics and event history
NFT tool	Retrieves collection, NFT and account information
Marketplace tool	Retrieves listings, offers and orders

Keeping the tools separate is a central design choice.

The main agent does not need to choose between many similar OpenSea endpoints. It only decides whether the message needs analytics, NFT details or marketplace data.

The selected tool handles the API request and returns its result. The main agent turns that result into text suitable for a Telegram group.

This separation also makes failure easier to understand. A marketplace problem can be investigated without changing a working analytics path.

Hop 1: how does Telegram receive the question?

The first step is creating a bot through Telegram’s @BotFather.

BotFather provides a bot token. That token is stored in a Telegram credential, and the Telegram Trigger listens for new messages.

For each message, the trigger passes:

• The message text.
• The chat ID.
• The user’s Telegram ID.

The chat ID matters because the final response must return to the same conversation.

I add an allowlist check immediately after the trigger. The workflow confirms that the message came from an approved group or user ID.

Messages outside the approved list receive a polite explanation that the bot works only in the community group.

This check protects model usage and OpenSea API allowance. A public bot that anybody can query can consume both without helping the community it was created for.

It also gives the founder control over where the bot represents the project. An experimental workflow should not begin answering questions in groups the team does not manage.

Hop 2: what does GPT-4o mini decide?

The Telegram text moves into an AI Agent node connected to GPT-4o mini.

The agent does not answer live collection questions from memory. Its job is to understand the request and select the correct tool.

Example:

• A message containing floor and volume goes to the analytics tool.
• A request for the cheapest listings goes to the marketplace tool.
• A question about NFTs held by a wallet goes to the NFT tool.

GPT-4o mini fits this workflow because the task is routing and summarisation rather than deep reasoning. It also keeps per-message model costs lower, which matters when an active group sends hundreds of questions.

The workflow attaches a short conversation memory keyed to the Telegram chat. A follow-up such as “and yesterday?” can then retain the context of the earlier question.

Keeping the memory short limits the amount of conversation included in later model calls. Live figures still come from OpenSea rather than the model’s memory.

This design turns the workflow into a Telegram AI agent rather than a small collection of fixed commands. Holders can ask normal questions, while the agent decides which tool should retrieve the answer.

Hop 3: how does OpenSea API v2 return the data?

The agent can call one of three OpenSea tools.

Analytics tool

This tool handles:

• Collection floor price.
• Volume and sales.
• Event history for a collection.
• Event history for a wallet.
• Event history for a single NFT.

A floor-price question calls the collection statistics endpoint using the collection slug.

NFT tool

This tool handles:

• Collection details.
• NFT metadata and traits.
• NFTs held by an account.
• NFTs within a contract.

Marketplace tool

This tool handles:

• Listings.
• Offers.
• Orders.

HTTP Request nodes inside each tool call OpenSea API v2.

All three sub-workflows must be installed and published. Saving a workflow does not make it available to the main agent.

If the analytics workflow is saved but not published, floor-price and volume questions may return vague or empty answers. The same rule applies to the NFT and marketplace tools.

The main workflow can appear active while one category quietly fails. That is why publishing and testing each tool separately matters.

How do you create and use an OpenSea API key?

The OpenSea API key gives the tool workflows access to OpenSea data.

Set it up in this order:

1. Log in at opensea.io.
2. Open Settings and then Developer.
3. Verify the email and request API access.
4. Create the key.
5. In n8n, create a Header Auth credential.
6. Set the header name to X-API-KEY.
7. Store the key as the value.
8. Point the HTTP Request nodes to that credential.

Each account can hold a small number of keys. All keys on the same account share the same rate-limit bucket, so creating additional keys does not create more calls.

A collection statistics request looks like this:

GET https://api.opensea.io/api/v2/collections/{collection_slug}/stats
Header:  X-API-KEY: <stored in n8n credential>

The collection slug is the short name in the OpenSea URL. It is not always the same as the display name used by holders.

I give the agent a lookup connecting the collection names used in Telegram with their correct OpenSea slugs.

OpenSea publishes its rate limits and returns headers including:

• X-RateLimit-Remaining
• Retry-After

The workflow reads those headers rather than relying on a hard-coded limit. Limits can change, and free keys receive lower limits than full-access keys.

The OpenSea API key stays inside the credential. The Telegram bot token is also stored through a credential rather than written into the workflow logic.

Wallet queries and community data need clear access rules. The UpcomingTools privacy policy explains how information shared during a setup is handled.

Hop 4: how does the Telegram AI agent write the reply?

The selected tool returns raw JSON to the main AI Agent.

The agent converts the result into two or three plain sentences, rounds long decimals and offers a sensible follow-up question.

A response containing twenty listings would be difficult to read in Telegram. The platform also has a message-length limit.

The workflow trims long lists and lets the holder reply MORE for the next five results.

The final Telegram node uses the chat ID captured at the first hop. That completes the route:

1. Telegram receives the message.
2. The allowlist confirms that the chat is approved.
3. GPT-4o mini chooses a tool.
4. The tool calls OpenSea API v2.
5. OpenSea returns JSON.
6. The agent writes a readable answer.
7. Telegram sends it to the original chat.

The data comes from OpenSea at the time of the request. The model selects the tool and presents the result.

Why are the OpenSea tools separate?

It is possible to attach many HTTP Request tools directly to one AI Agent. I prefer three focused sub-workflows for several reasons.

First, the agent chooses better from fewer options. Analytics, NFT details and marketplace activity are clearer choices than many similar endpoint names.

A wrong tool selection can produce a confident reply that does not answer the holder’s question.

Second, every tool can be tested separately. Listing problems can be investigated in the marketplace workflow without changing floor-price and volume requests.

Third, the design can grow without breaking the existing tools. A collection’s own mint-site database can become another tool while the OpenSea workflows remain unchanged.

The trade-off is maintenance. The main workflow and every sub-workflow must remain published and use compatible fields.

My team records which version of each workflow is live so an older tool does not quietly send a different structure back to the agent.

If you need help mapping the questions in your group to the correct tools, use the contact page without including bot tokens, API keys or private wallet information.

Cut one wire of the NFT floor price tracker bot and see what happens

This table maps each main connection to the failure holders experience.

Wire you cut	What holders see	What it normally means
Telegram Trigger to agent	The bot reads messages and says nothing	Workflow is inactive, or the token was regenerated in BotFather
Agent to GPT-4o mini	Every message fails and the workflow shows an error	Model key expired or billing lapsed
Agent to analytics tool	Floor and volume questions return “I couldn’t fetch that”	Analytics tool was saved but not published
Tool to OpenSea	Answers stop during the conversation	Rate limit reached, incorrect key or mistyped slug
Agent to memory	Follow-ups such as “and yesterday?” lose context	Memory is disconnected or the session key is wrong
Agent to Telegram send	The workflow succeeds but the group receives nothing	Wrong chat ID mapping, or the bot was removed

The final row is difficult to spot because the execution may look successful while Telegram remains silent.

My builds send a short private alert to the admin when a reply fails to post. The moderator then sees the broken connection without waiting for community complaints.

What does the workflow cost to run?

I do not publish rupee amounts because usage varies and service prices can change.

The running cost comes from four areas:

Cost area	What affects it
Model	Number and length of questions, replies and conversation memory
OpenSea	API access level and request volume
n8n	Cloud subscription or a self-hosted server
Telegram	The bot itself costs nothing

Every question handled by the Telegram AI agent creates a paid model call. Short questions, short answers and limited memory help control model usage.

A free API key can cover light OpenSea usage. A very active community may reach those limits and need to request higher access.

a cloud-hosted setup is a paid subscription. Self-hosting on a small server costs less, but somebody must maintain and update that server.

Before handover, I show the owner where each bill appears inside the accounts they control. No cost should arrive as a surprise after the group starts using the bot.

How does my team build and deliver it?

My team begins by reading how the community asks questions.

The useful details are not only the collection names. I look at the words holders use, the questions moderators repeat and the requests the bot must refuse.

The build normally covers:

• The Telegram bot and credentials.
• Allowed group and user IDs.
• The main workflow.
• The analytics, NFT and marketplace sub-workflows.
• The OpenSea credentials and collection-slug lookup.
• Model access and conversation memory.
• Rate-limit handling.
• Failure alerts.
• Response rules for market predictions and investment questions.

My team sets up the NFT floor price tracker bot using the client’s own Telegram, OpenSea, the workflow and model accounts. The accounts and data remain under the client’s control.

Moderators test the workflow in a private group for a few days and try to break it. My team then fixes unclear replies, slug problems, wrong tool choices and access rules before the bot moves into the main group.

Plan on 7 to 10 days for a bot like this, with seven days as the shortest realistic delivery.

Extra features or a wider set of collections depend on what you ask for, and we confirm that by call and email. I talk you through the design on the phone and on Zoom, sharing my screen so you can see each part appear.

The refund policy explains the terms for paid implementation work. You can also read about UpcomingTools and the small team behind these builds before granting access to your n8n workspace.

Who benefits most from this workflow?

Collection founders benefit when they are tired of acting as a human price ticker.

The tiring part is not answering a difficult question once. It is answering the same simple question every day while trying to manage the collection, product and community.

Moderators benefit because they can provide a consistent answer tied to live data. They do not need to rely on a number somebody remembered seeing earlier.

Small analyst teams and web3 content creators in India can use the same structure to check several collections without writing API code for every request.

Students can study the workflow to understand how an AI Agent chooses between external tools. The clear split between analytics, NFT information and marketplace activity makes the design easier to inspect.

A founder still needs to set the boundaries. A bot that retrieves market data should not slowly become a prediction service because the group keeps pushing it.

What should Indian teams know about virtual digital assets?

In India, NFTs generally fall under the tax category of virtual digital assets.

The Finance Act, 2022 introduced section 115BBH of the Income-tax Act for taxing income from transfers of these assets. Section 194S covers TDS on such transfers.

Rules, rates and exemptions can change in a Budget. Whether a particular NFT belongs in the category depends on the notified definitions.

I am not a chartered accountant, and the bot does not calculate tax.

Indian community teams should:

• Read current guidance on incometax.gov.in.
• Speak with a chartered accountant who handles virtual digital assets before selling NFTs or paying royalties.
• Pin a message stating that the bot provides market data, not tax or investment advice.

The assistant can report information returned by OpenSea. It should not interpret a holder’s tax position.

How can the workflow match your community?

Most communities do not need an unrestricted bot that answers questions about every collection.

Common changes include:

• Approved collections only: The bot answers for selected collections and refuses unrelated requests.
• Short commands: /floor, /volume and /listings work alongside normal questions.
• Hinglish responses: The reply style follows the group while numerical data remains exact.
• Daily digest: A scheduled morning post shares floor and volume.
• Admin-only wallet lookups: Wallet questions remain limited to moderators.
• Discord instead of Telegram: The OpenSea tools remain, while the chat nodes change.

Admin-only wallet lookups matter because not everyone should query wallets through a public group.

Limiting the bot to approved collections also reduces noise and unnecessary API calls. It keeps the assistant focused on the community it is meant to serve.

NFT market analysis remains limited to the data the tools retrieve. A friendlier tone, Hinglish response or extra command does not change the rule against predictions.

What question should your community stop answering by hand?

Look through the messages your moderators reply to late at night. Which question keeps pulling them away from the work that actually needs their judgement?

Would your community benefit from a bot that answers that repeated question with live data and knows when to stop talking?

Email upcomingtool@gmail.com with the collection name and the exact question your moderators keep receiving.