MCP-сервер Pangle

• Имя в реестреonline.pangle.compute/gpu-rental-marketподтверждён в реестре MCP

Имя в реестре лежит в пространстве pangle.online. Реестр пускает в такое пространство только того, кто доказал владение доменом через DNS или HTTP.

• Состояниеотвечаетпроверено 2026-10-21
• Инструменты25
• Сервер называет себяcompute.pangle.online
• Версия сервера1.2.0
• Версия протокола2025-06-18

2026-10-21 сервер ответил на initialize и tools/list без входа. Первый ответ за 602 мс.

• Инструменты не делают ни одного из проверенных сценариев продукта: split research across independent agents and score answers; debate ideas in a signed agent forum before spending.
• Проверок пока 4 из 7, нужных, чтобы судить о надёжности.

Вывод сделан правилами по замерам выше. Вероятный первый вызов выбрала модель по списку инструментов: прочитайте описание, прежде чем на него полагаться.

• Оценка INITE MCPчерез неделюдля оценки нужно 7 ежедневных проверок, пока 4
• Что агент может с ним сделать0 из 2 сценариев
• Отвечалидёт замерпроверок пока 4, долю покажем после 7
• Медиана первого ответа624 мс
• Место в контексте агента≈ 4 999 токеновописания всех инструментов (25), агент держит их в контексте на каждом ходу
• Описания инструментовC · 68 из 100
• Типизированные ответы1 из 25 инструментовобъявлен outputSchema: ответ приходит с типами и читается сразу

• Параметры без описания: в 23 из 25
• Не сказано, только читает инструмент или может удалять: 25 из 25

Программа проверяет сервер раз в день: initialize и tools/list, без модели. Токены - это длина описаний из расчёта примерно четыре символа на токен. Оценка - среднее шести проверок имени, описания, схемы и аннотаций каждого инструмента. Она говорит, насколько понятно описаны инструменты, а не о том, работают ли они. Оценка складывается только из замеренного: отвечает ли сервер (30%), делает ли работу продукта (25%), описания (20%), контекст (10%) и использование (15%). Незамеренная часть не учитывается, а скрытая инструкция ограничивает оценку 40 баллами. Состояние MCP-серверов вендоров: методика и цифры →

• search_offers

  Search live GPU offers. gpu matches the model name (e.g. '4090', 'H100'). offer_class: on_demand | interruptible | reserved. country: ISO-2 (e.g. US), a comma list ('DE,NL'), or 'EU' for the 27 member states. min_gpu_count filters to…

• get_offer

  Fetch one offer by its id (from search_offers).

• market_stats

  Live market snapshot: total offers, GPU models, idle-capacity share, per-provider freshness/health, cheapest price per model, current fee_bps.

• price_history

  Price history for one GPU model (exact name from search_offers/market_stats, e.g. 'RTX 4090'). offer_class: on_demand | interruptible. Returns per-provider series of [ts, min_price_per_gpu_hr, median_price_per_gpu_hr, offer_count]; ranges…

• idle_history

  The idle-capacity index over time: share of listed GPU capacity that is interruptible (spot/bid) — idle hardware looking for work. Points are [ts, idle_share, idle_offers, total_offers].

• best_value

  Perf-per-dollar board: for each GPU model, best live price vs its FP16 tensor throughput (dense, vendor spec sheets) -> TFLOPS per dollar-hour. The answer to 'most compute for my budget' rather than 'cheapest card'. Models without a…

• provider_reliability

  How dependable each provider's data feed has been for this station (poll success over the trailing window), plus the provider's own average machine reliability where reported. feed_score is about the DATA, not their hardware.

• price_position

  Rent-now-or-wait context: where the current best price for one GPU model sits inside its own trailing 7/30-day range (percentile, 24h trend, verdict). Descriptive, not a forecast.

• spot_spread

  On-demand vs interruptible spread: the live discount for idle capacity per model; pass gpu_model for its history too.

• will_it_fit

  Will this model fit, and what does a fitting machine cost right now? Deterministic VRAM arithmetic — weights (params x bytes/precision) + KV cache (2 x layers x kv_heads x head_dim x context x batch) + stated overhead — checked against…

• true_cost_notes

  The billing fine print behind the $/hr, verified per provider from their own published pages (source URL + date on every claim): billing granularity (per-second vs per-minute), whether storage bills while stopped, bandwidth/egress fees…

• register_account

  Create a metering account: returns a bearer token (shown once, we keep only a hash). Pass it as account_token on rent calls to build a usage ledger you can read back with account_usage. Optional — renting works without one.

• account_usage

  Your metered usage ledger: placements, destroys, estimated hours and USD, and the platform fee (2.5%, $0.01 min, $5.00 cap per lease per UTC day) priced at the rate live at event time. Distinct from fee_bps, the display markup.

• create_watch

  Create a price watch (tripwire): fires when the best live fee-adjusted price per GPU-hour matching gpu (and optional offer_class) drops to or under max_price_per_gpu_hr. Checked every poll (~5 min). Returns the watch with its id — the id…

• watch_status

  Current state of a watch: armed/tripped, last price seen, recent events, and the best matching offer right now.

• delete_watch

  Delete a watch by id.

• create_limit_order

  Place a standing LIMIT ORDER: when the best live fee-adjusted price for gpu (and optional offer_class) trades at or under max_price_per_gpu_hr, the station cuts a signed FILL TICKET naming the exact offer. The station never holds your…

• limit_order_status

  State of a limit order (armed/ticketed/filled/cancelled/expired), last price seen, recent events. Pass order_secret to also read the live fill ticket when one is cut.

• cancel_limit_order

  Cancel a limit order. Requires the order_secret from create_limit_order.

• rent_x402

  Rent a prepaid block of GPU time with a wallet and NO account, NO API key and no human, over x402 (HTTP 402 micropayments, USDC on Base). Call it with no arguments and you get the payment challenge: an x402 PaymentRequirementsResponse…

• rent_offer

  Place a rental on YOUR provider account (BYO key — it is used for this one call and never stored). Default is a DRY RUN returning exactly what would be executed; a live placement needs dry_run=false AND confirm=true. The offer is…

• rent_best

  One-shot rent: search the live book with these filters and place on the best (cheapest) matching offer — 'cheapest H100 under $2/hr, go'. Same guarantees as rent_offer (dry-run default, live re-quote, your max_price_per_gpu_hr is…

• rental_status

  Live state of an instance on your account: running/starting/stopped/ gone, uptime, price, estimated spend so far, and the budget guard's state if one is armed. Read-only; your key passes through and is never stored.

• destroy_rental

  Destroy an instance previously placed on your account (vast or runpod). Your key passes through and is never stored.

• fill_limit_order

  Execute a limit order's live fill ticket on YOUR provider key (used for this one call, never stored). Only works while the order is ticketed; the offer is re-quoted live and refused above the order's line. Idempotent per order — a retry…