Undertow MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| agent_access_statusA | Identity, current entitlement, daily usage and access instructions; explains missing/refused subscriber tools. |
| crypto_exit_checkB | Compare four BTC sell sizes by rung/bracket. Withholds stale/future, uncovered or unconverted venues. Fees excluded; no trade clearance. |
| crypto_funding_costB | Caller-input perpetual funding, fees, net carry and linear break-even. Positive funding: longs pay shorts. Constant-rate scenario, not forecast. |
| crypto_portfolio_stressB | Caller-input linear shocks for 20 exposures: P&L, equity, asset/venue concentration. Symbols are scenario labels. No liquidation/probability model. |
| depth_episodesB | Realized BTC depth-collapse onset, trough, recovery under preregistered rules. Zero is distinct from missing. Subscriber exit_desk_full adds ETH. |
| exit_costA | BTC nearest-rung sell estimates, interpolated bid-depth bands; fees excluded, not executable. crypto_exit_check compares sizes/freshness. |
| gold_cash_realisationA | Calculate gold sale proceeds and cash gates from caller inputs. Decimal strings: kg, fineness (0,1], USD/troy-oz bid, USD fees, currency/USD FX (USD=1). Use aware ISO clocks. Returns arithmetic and withheld-cash reasons; no fetch, storage, execution or verified balance. |
| latest_articleA | Read the latest published editorial: title, full text, URL, clock and quality receipt. No arguments. Missing text or failed receipt returns an error; no generated prose. Use liquidity_tiers for board data. |
| liquidity_tiersB | Nine segment tiers, Seiche funding, measure counts and replay/withholding reasons. PARTIAL means missing history or unpassed replay, never calm. |
| market_corpus_contextB | Inspect corpus coverage, rights and clocks. Summary takes no pagination; other views return records/total/next_offset. Retrieval age is not observation freshness. Use liquidity_tiers for market readings. |
| research_networkB | Find cross-product source metadata, clocks, rights and next steps. Follow next_offset; unavailable flags retrieval, contract or size failure. Use market_corpus_context for local corpus counts. No shared score or authority. |
| sealed_recordB | Forward calls sealed/signed before outcomes: hits, misses, decided counts, ledger root. Report hits and misses together. |
| trade_safety_exit_contextB | Exact-rung BTC/USD observe/paper context: request hash, PIT source/rights, conversion, bid bands, six venue clocks, startup SHA. No live mode or clearance. |
| unwind_watchA | Watched-filer headlines: effective positions, days-to-exit brackets, blind spots. No live execution model. unwind_stress has full scenario grids. |
| venue_concentrationA | BTC two-sided depth: HHI, effective/top venue and venue depth. Subscriber exit_desk_full adds ETH and venue failure. |
| venue_price_reconciliationB | BTC/ETH consensus: depth/squared-half-spread weights, deepest-venue blindness, concentration, dislocations. Experimental; no execution authority. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| can_this_book_exit | Unwind stress for the watched books: door width versus position size, margin clock, and what changed this quarter. |
| exit_cost_check | Venue-by-venue exit cost for a position size, with concentration and consensus-price context and the caveats stated. |
| market_liquidity_briefing | The full board read: tier per segment, realized depth episodes, and the sealed forward-calls record with misses quoted. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 16 tools
Most tools target distinct data products such as funding cost, gold sale proceeds, editorial retrieval, and access status. However, the BTC exit surface is split across exit_cost, crypto_exit_check, and trade_safety_exit_context, while venue_concentration, venue_price_reconciliation, depth_episodes, and liquidity_tiers overlap in market-microstructure readings. Descriptions help distinguish them, but an agent must read closely to avoid misselection.
All tool names use lower_snake_case consistently and are readable. The pattern is mostly noun-phrase/domain-prefix rather than verb_noun, with minor semantic inconsistencies such as asset-prefixed versus context-suffixed names, but there is no chaotic mixing of conventions.
16 tools is reasonable for a multi-domain research and analytics server spanning crypto, gold, editorial, access, and corpus workflows. It is slightly above the ideal 3-15 range, but each tool appears to cover a specialized surface rather than duplicating a simple operation.
The surface covers a broad range of research and risk workflows including exits, funding, stress, depth, venue reconciliation, corpus context, and editorial retrieval. Gaps are mostly intentional or subscription-gated, such as no live execution and limited ETH coverage in some tools, which agents can work around.