> For the complete documentation index, see [llms.txt](https://polydash.gitbook.io/polydash-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://polydash.gitbook.io/polydash-docs/agents-and-api/ai-agents.md).

# AI agents

Polypad's agents are open source, built on [Swarms](https://swarms.world), and non-custodial. They read on-chain data. They do not sign or send transactions.

| Agent          | What it does                                                                                                     | Tools and limits                                                                                                                                         |
| -------------- | ---------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **PolyCTO**    | An AI CTO for a Polypad launch. Four roles: community manager, marketing expert, token analyst, security analyst | Token analysis, holder and flow analysis, a heuristic source scan, and a scam-message checker. The scan is not a review                                  |
| **Poly Agent** | A read-only staking analyst for Polypad pools                                                                    | Lists pools, reports totals, timing, token-denominated reward rates, and recent stake and withdraw flows                                                 |
| **PolyForge**  | A post-launch builder for staking                                                                                | Builds staking-pool transactions, drafts milestone posts from on-chain data, and opens a token's CTO. You review every field and sign in your own wallet |

## PolyCTO

Four roles share one keyword router:

| Role              | Tasks                                                                               |
| ----------------- | ----------------------------------------------------------------------------------- |
| Community manager | Telegram welcomes, FAQ answers, engagement posts, spam and scam spotting            |
| Marketing expert  | Launch plans, X and Telegram drafts, campaign calendars                             |
| Token analyst     | Name, symbol, decimals, supply, owner, top holders, concentration, recent transfers |
| Security analyst  | Heuristic scan of Sourcify-verified source, plus on-chain owner and proxy checks    |

In Telegram mode the bot welcomes new members with official links, answers when mentioned, and scores every message for scam patterns. It deletes nothing and bans nobody by default. Suspected scams are logged as suggestions. Auto-moderation is opt-in and never bans.

The security scan flags patterns such as privileged mint, blacklist, pausable, fee setters, trading switches, upgradeable proxies, owner withdraw, and whether `owner()` is renounced. It skips library files. **It is a heuristic scan, not a review.** Regex checks can miss obfuscated logic or flag harmless code.

## Poly Agent

Poly Agent monitors staking pools on Polypad. A reward rate it reports is a snapshot in tokens at the current stake. It is not a promised return. Use cases:

* Pool flow alerts for large deposits and withdrawals.
* Pool lookup for total staked, reward rate, and time remaining.
* A list of open pools and their terms.

## Run an agent yourself

Both agents are in the [repository](https://github.com/polydash-labs/Polydash/tree/main/agents). They need Python 3.10 or later.

```bash
cd agents/poly-agent
pip install -r requirements.txt
cp .env.example .env
python poly_agent.py --tools      # raw on-chain pool data, no LLM
python poly_agent.py --watch      # live inflow and outflow alerts
```

```bash
cd agents/polycto
pip install -r requirements.txt
cp .env.example .env
python polycto.py --analyze 0xTOKEN
```

Never commit your `.env` file.

## Guardrails

Agent prompts forbid price predictions, return promises, and investment recommendations. They never ask for seed phrases or private keys, and they only share configured official links.

## Risks

Agents can make mistakes. Their replies are information, not advice. Verify anything important yourself.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://polydash.gitbook.io/polydash-docs/agents-and-api/ai-agents.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
