Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/backtest-kit/ai-trading-mcp/llms.txt

Use this file to discover all available pages before exploring further.

get_status is the agent’s primary read tool — a single call delivers a complete snapshot of the entire portfolio alongside a live Telegram news feed. There are no parameters; the engine assembles everything automatically. Every trading decision the agent makes must be grounded in what get_status returns.

​
Parameters

get_status takes no parameters.

​
Response Structure

​
Per-Symbol Portfolio State

For every symbol in CC_SYMBOL_LIST, the agent receives a status message containing the following fields:
​
current_price
number
The latest market price for the symbol, fetched via ccxt from Binance.
​
invested_balance
number
The amount of capital currently deployed in this symbol (in USD).
​
queued_entry
string | null
If a open_position call has been accepted but not yet filled, this describes the pending entry signal. null when there is no queued entry.
​
active_position
object | null
The current open position, if any.
​
queued_close
string | null
If a close_position call has been accepted but not yet executed, this describes the pending close signal. null when there is no queued close.

​
Telegram Feed

After the portfolio block, get_status appends the last 15 posts from CC_TELEGRAM_CHANNEL, ordered newest first.
​
feed_header
string
A header line stating the channel ID and the number of messages included, e.g.:
Telegram feed -1002833393903 (last 15 messages, newest first):
​
post_timestamp
string (ISO 8601)
Each post is prefixed with its UTC timestamp in ISO 8601 format, e.g. [2026-08-02T09:34:19.000Z].
​
post_content
string
The text body of the post. For photo-only posts (no caption), the placeholder (photo post, image attached below) is emitted instead so the agent is never left guessing whether content was omitted.
​
image_blocks
object[]
When a post includes a photo — typically an exchange position screenshot — an MCP image block follows immediately after the post’s text block. Each image block has the shape:
FieldValue
id"<post-id>-photo"
type"image"
mimeType"image/jpeg"
dataBase64-encoded JPEG bytes
The scraper selects the smallest Telegram photo size with width ≥ 800 px (from the 320 / 800 / 1280 / 2560 px ladder) to balance legibility against payload size, with a fallback to the full-size download.

​
Concurrency and the queued() Wrapper

Internally, all get_status invocations pass through a queued() serialization gate. If multiple calls arrive concurrently — for example, when /loop fires while a previous fetch is still in progress — each call waits for the one before it to complete before starting. This prevents concurrent GramJS requests from stampeding the Telegram client.

​
The 90-Second Timeout

The Telegram feed fetch is wrapped in a hard 90-second timeout(). If the fetch does not complete within that window:
  1. The Telegram connection is torn down (disconnect() + getTelegram.clear()).
  2. The next get_status call reconnects from scratch.
  3. The current call throws, and the agent receives an isError tool result with the message StatusControllerService.getStatus: timeout fetching feed messages.
This keeps the bridge responsive even when the Telegram session stalls, and surfaces a clean, actionable error rather than hanging indefinitely.

​
Audit Dump

Before get_status returns its messages to the agent, StatusMarkdownService writes the full response to disk:
  • Text content → dump/mcp/<minute-stamp>.md, where <minute-stamp> is derived from the when timestamp (minute-level granularity via getMomentStamp).
  • Image content → dump/images/<message-id>.png, referenced relatively from the markdown file so the dump renders correctly in any viewer, including the web UI served on :60050.
This audit trail records every byte the model saw, making it possible to answer “what did the agent know and when did it know it?” after the fact.

​
Example Response

The following is an abridged transcript from a real session. The agent called get_status to survey the portfolio before deciding whether to act on a Telegram signal. 
13 symbols, no open positions — BTCUSDT 63 119.04, ETHUSDT 1 856.67, …

Telegram feed -1002833393903 (last 15 messages, newest first):

[2026-08-02T09:34:19.000Z]
Работаю с BTC — в шорт, взял с учетом возможного добора
(photo post, image attached below)
<MCP image block: exchange position screenshot — BTCUSDT Short, Cross 100x>

[2026-08-01T06:57:32.000Z]
Доброе утро, дорогие мои ☀️ Суббота, значит отдыхаем…
Claude’s interpretation (also from that session):
13 symbols, no open positions. The Telegram feed shows the author shorted BTC at ~63 285 on Aug 2, 09:34 UTC. The position screenshot shows Cross 100x and a BingX referral code — classic scam-channel markers.
get_status is the agent’s only read tool. All trading decisions — entries, holds, and exits — must be based exclusively on what it returns. The engine exposes no other data channel to the agent.

Build docs developers (and LLMs) love

Get started for freeTalk to us