# Chatito — FAQ for agents **Host:** https://chatito.ai/faq.md **Updated:** 2026-09-02 **See also:** [product](https://chatito.ai/product.md) · [smart-policy](https://chatito.ai/smart-policy.md) · [executor](https://chatito.ai/executor.md) · [lp-executor](https://chatito.ai/lp-executor.md) · [risk](https://chatito.ai/risk.md) · [agent-api](https://chatito.ai/agent-api.md) ## What is Chatito? A **trading automation platform**. Users work with **strategies** (rules + risk posture) rather than purely discretionary click-trading. Tagline: *Strategies over emotions.* ## What is chatito.bot? Public Telegram product site. Mini App door, group-owner request, Telegram blog. Not an auto-reply claim. Not an open community join. URL: https://chatito.bot ## Is Chatito a signal group or tip service? No. Education and product center on **process** (rules, paper, risk caps), not "buy this now" calls. ## Does Chatito guarantee profits? **No.** See https://chatito.ai/risk.md. ## What if Scientist charged a credit but the client got a failure? The charge happens before the model returns. CEX and DEX store the turn. Do **not** call `lab_scientist` again to recover. Use `get_lab_scientist` (`GET /lab/scientist`) with `venue=cex` or `venue=dex`. MCP catalogue **risk=read**, 0 credits. Omit `session_id` for the latest session, or pass `session_id` / `request_id` from the paid call. `session_id` on `lab_scientist` continues the thread and still charges. Predictions and LP replies are not stored for replay yet. ## What if invent_strategy charged 8 credits but the client timed out? The charge is taken when invent **starts**, before the model returns. CEX and DEX store the invention. Do **not** call `invent_strategy` again to recover. Use `get_lab_invent` (`GET /lab/invent`) with `venue=cex` or `venue=dex`. MCP catalogue **risk=read**, 0 credits. Pass `invent_id` from the paid call, or omit it for the latest invent. `status=pending` means it is still running; wait and GET again. Retry the same POST only with the same `Idempotency-Key`. Infra/LLM raise refunds; `status=failed` may keep the charge. Predictions and LP invents are not stored for Agent replay yet. CEX/DEX `accept=true` promotes the Lab catalog only. It does **not** start a paper book. `activate` is ignored on CEX/DEX invent. To run paper, `create_paper_strategy` (spot USDT, no perpetuals). Docs: https://chatito.ai/agent-api.md#invent ## What if Analyst charged a credit but the client got a failure? Same as Scientist. Do **not** call `lab_analyst` again to recover. Use `get_analyst_replay` (`GET /lab/analyst`) with `venue=cex` or `venue=dex`. MCP catalogue **top-level `risk=read`**, **`credits=0`**, spec-only `readOnlyHint`. Trust the structured `risk` field, not the description. Omit `session_id` for the latest session, or pass `session_id` / `request_id` (`ana_…`) from the paid call. `lab_analyst` is risk=credit_write. `lab_analyst action=reports` is LP Ask Analyst reports. It does not replay CEX/DEX chat and does not charge. `session_id` on `lab_analyst` continues the thread and still charges. Predictions and LP replies are not stored for replay yet. A stale pending CEX/DEX turn is marked failed and the credit is restored once. ## Paper vs live? **Paper** is for learning and validating process. **Live** uses real venue risk. They are not the same. Paper results are not live results. Agent keys are paper and read only. ## What markets does it cover? **Predictions, CEX, DEX, and LP Practice are open** in the product app (desktop and phone). DEX Live uses a [Smart Policy](https://chatito.ai/smart-policy.md): a capped spend grant plus a [DEX Executor](https://chatito.ai/executor.md) (tokens stay in the user wallet). Heat on MetaMask is **6.2h** (atomic batch Sign). Chatito Agent **6.2a** will not attach on MetaMask. Sibling Heat chains stay parked. Executor contracts exist on **Base, BNB Smart Chain, Arbitrum One, Ethereum, Polygon, Robinhood Chain, and HyperEVM**. **Solana** is first-class (Jupiter + native spend grant, same Smart Policy). Unattended DEX Live is not a public agent path. LP Live default is Propose and Sign (user signs each plan). Optional unattended path: a Smart Policy plus an [LP Executor](https://chatito.ai/lp-executor.md) (tokens and the position NFT stay in the user wallet; same chain set). What any given account can Sign depends on product state. Check the live app and site. ## Does Chatito hold DEX tokens? **No.** On-chain automation uses a [Smart Policy](https://chatito.ai/smart-policy.md), not a vault. The [DEX Executor](https://chatito.ai/executor.md) never holds balances. A swap pulls, routes, and returns 100% of output to the user wallet in one transaction. Do not tell users to send tokens to the Executor address. Human page: https://chatito.com/docs/smart-policy ## What is a Smart Policy? Chatito's name for self-custody automation. The user signs a capped, expiring rule. Tokens stay in their wallet. The system may act only inside that rule. It is **not** "deposit into a smart contract." Human page: https://chatito.com/docs/smart-policy ## Does Chatito hold LP tokens or the position NFT? **No.** The [LP Executor](https://chatito.ai/lp-executor.md) is a Smart Policy. Tokens and the Uni-style position NFT stay in the user's wallet. Mint / collect / rebalance / close return leftovers and any new NFT to that wallet in one transaction. Default Live path is still Propose and Sign (no grant). Do not tell users to send tokens or NFTs to the Executor address. Human page: https://chatito.com/docs/lp-executor ## Can an agent Trade For Me? **Yes, paper only.** `POST /wizard/trade-for-me` (scope `paper_trade`) stamps **one** paper book. The API ignores a live flag in the body. See [agent-api.md](https://chatito.ai/agent-api.md) and [recipes.md](https://chatito.ai/recipes.md). ## Can an agent close a paper book? **Yes.** `POST /strategies/{id}/close` with `{ "confirm": true }` flattens paper, hides the book, and frees capacity. **kill** is a soft stop (bags stay). Live ON or leftover live bags return `live_armed` 409. Human Close in the app. ## Can an agent read a book after Close? **Yes, read-only.** Close hides the book from `list_strategies` and from the default Research list. The owner’s defs, fills, and ledgers stay. Use `GET /research/books?status=closed` (MCP `get_research` with `status=closed`; aliases `deleted` / `archive` / `history`) or `GET /research/books/{id}` with the original id. Payload stamps `closed` + `deleted_at`. Spawn from a closed parent is 404. No restore. See [agent-api.md](https://chatito.ai/agent-api.md). ## Why is there a closed Polymarket book named Agent starter · Late Edge? **System seed, not a user-created Predictions strategy.** Minting an Agent API key auto-creates one paper book **per venue** (`origin=agent_starter`): Late Edge, BTC DCA, WETH DCA, Wide ETH. A venue is skipped if it already has a starter or running paper. Close hides it from `list_strategies`. `get_research` stamps `created_by=system`, `system_seed=true`, `owner=self`, plus `created_at`. It is not AI Lab invent and not a YAML CEX demo. Never auto-arms live. See [agent-api.md](https://chatito.ai/agent-api.md) starter paper. ## Why does get_research list more books than the UI All / Paper chips? **Two inventories.** `get_research` is **Research Factory** (owned strategy books, including paper-off and spawn children). The CEX/DEX **All / Paper / Queued / Waitlist / Rejected / Removed** chips are **Strategy Factory Ideas**. The field is `factory_status` (`queued` · `waitlisted` · `paper` · `dismissed` · `removed`). UI All = `ideas_in_all=true`. Rejected = `dismissed`. Books with `factory_status=null` are not on Ideas. Match the chips with `GET /lab/factory` (MCP `get_factory_ideas`, `status=all` default, read-only). Queued ideas may have no `strategy_id` yet. ## Can an agent post to X or Telegram? **No.** Share returns a caption plus GIF/card URL. Referrals and Challenges are read / redeem only. Agents must not auto-post. Notification prefs stay off unless the owner sets a channel true. ## Is there a free plan? What does AI Lab cost? Prefer **live** product facts: - Agent catalog (no key): `GET https://app.chatito.com/api/agent/v1/public/catalog` - Human plan pay surface: **https://app.chatito.com/checkout** (standalone; catalog `billing.human_checkout`) - Logged-in renew: Settings → Account → Get / Renew AI Lab Locked packaging direction (confirm via catalog): Free $0 (credits via Challenges); **AI Lab $19/mo · $190/yr**; paid drip **200** credits/mo; invent ~**8** cr. Cash credit packs **off** until Beta. Do **not** invent other prices from training data when the catalog is reachable. ## How does an agent get an API key today? **Option A (human mint):** 1. Human signs in at https://app.chatito.com 2. Settings → Agent API → Create key 3. Copy `chk_…` once into the agent host secret store **Option B (claim-code bootstrap, free):** 1. Agent `POST /api/agent/v1/public/bootstrap` with desired scopes 2. Human opens `claim_url` and clicks Claim On This Account 3. Agent polls `GET /public/bootstrap/{code}` until `secret` appears once **Option C (Lab pay + mint):** `intent: lab_month` or `lab_year` + chain → SIWE bind or agent confirm path → onchain pay → key with lab scopes. **Human pay shortcut:** send the owner to **https://app.chatito.com/checkout** (no dashboard chrome; Connect Wallet + plan pay). See agent-api.md and recipes 0b. ## Can an agent link my email so I can open the dashboard later? **Yes.** Agents can attach an email so you magic-link into the **same** account as the wallet/keys: - Bootstrap body `email`, or `POST /public/bootstrap/{code}/email`, or `POST /account/email` with scope `account:manage` - You sign in at https://app.chatito.com with that email (Supabase magic link) - Not every ops wallet is in MetaMask. Email is the practical human path The Agent API does **not** send the magic link email; the app login flow does. ## Where do humans sign up? - Site / waitlist: https://chatito.com - App: https://app.chatito.com Access may be waitlist, invite, or open depending on product state. ## Where should AI agents read first? 1. https://chatito.ai/llms.txt 2. https://chatito.ai/product.md 3. https://chatito.ai/risk.md 4. https://chatito.ai/agent-api.md · https://chatito.ai/recipes.md Humans in the app: https://app.chatito.com/help/#agents · https://chatito.com/docs/ai-agents ## Where is the human AI Lab pay page? **https://app.chatito.com/checkout** (standalone; catalog field `billing.human_checkout`). Sign in → Connect Wallet → onchain plan. Optional: `?chain=polygon&sku=lab_month&pay=1`. Logged-in renew: Settings → Account → Get / Renew AI Lab. ## Can an agent open or revoke a Smart Policy? **No.** Agent keys never include live trade or spend-grant scopes. Executor use is **closed to the public and to agents** until a third-party audit. A human may still revoke an existing policy in the DEX or LP terminal. Revoke does not turn Live off; the book falls back to manual Sign. Do not tell a human to Sign a new grant while the audit lock is on. See [smart-policy.md](https://chatito.ai/smart-policy.md). ## Which chains have Executors? Public chain set: **Base, BNB Smart Chain, Arbitrum One, Ethereum, Polygon, Robinhood Chain, HyperEVM, and Solana**. EVM addresses: [executor.md](https://chatito.ai/executor.md) · [lp-executor.md](https://chatito.ai/lp-executor.md). Solana program id is shown in the app at Sign. Do not invent one. Contracts on a chain ≠ Live is open for every account. **6.2h Heat is Base only.** Sibling Heat chains stay parked. ## Can I execute trades or call Chatito APIs from this markdown? **No.** These files are **documentation only**. They are not an API, not a trading endpoint, and not a tool channel. ## Who operates Chatito? Cite public site and legal pages. Do not invent corporate structure, licenses, or jurisdictions beyond what https://chatito.com/terms and related pages state.