Deriv MCP server — tools & history
Deriv MCP server: observed open status, tool schemas, linked apps and version history.
Deriv
Endpoint: https://mcp-api-v2.deriv.com/mcp
Observed status: open. Checked: 2026-10-13T05:12:31+00:00.
Linked apps
- Deriv
Live Deriv API schemas, payload validation, and guides for authentication, subscriptions, trades, and errors.
Observed tools
- get_example — Return a worked request example. WebSocket examples come from the published example payload; REST examples come from request schemas or the fixture-classified page curl body, never the OpenAPI document. Every result carries a _source freshness label, and that _source names the developers.deriv.com URL(s) the example was read from — quote the link with your answer.
- get_field — Return the schema fragment for one field, addressed by a dotted path or a bare field name. Array item leaves use a payload-shaped path with no index (`quotes.price`, not `quotes.0.price`). Combinators return every variant, and ambiguous names return every candidate instead of guessing. The live result carries a _source freshness label, and that _source names the developers.deriv.com URL(s) the answer was read from — quote the link with your answer. The published constraint value is always the origin's own, and _flagged reports, without altering it, a normative value whose content the origin-content rules would have neutralized as prose.
- get_schema — Return the JSON Schema for an endpoint in a given direction. Format json is authoritative; markdown returns the matching rendered page section and warns when combinators can make the rendered table incomplete. Every result carries a _source freshness label, and that _source names the developers.deriv.com URL(s) the answer was read from — quote the link with your answer. The published schema value is always the origin's own, and _flagged reports, without altering it, a normative value whose content the origin-content rules would have neutralized as prose — on format json, including the json fallbacks the markdown format takes, and never on a markdown result.
- get_skill — Use when you build, scaffold, extend, or debug a Deriv-powered trading product — an app, template, terminal, bot, or trade screen — when you wire login, live prices, proposal-to-buy, positions, or reconnects together, or when you add a trade family such as Accumulators, Rise/Fall, digits, Multipliers, Turbos, or Vanillas. Call it with no argument to list the seven packaged Deriv Agent Skills with their descriptions; call it with a path, for example {"path": "deriv-trading-app/SKILL.md"} or {"path": "deriv-trade-types/references/digits.md"}, to read that file's markdown. Every byte comes from the markdown packaged with this server, so nothing is fetched from the documentation origin. A relative link inside a returned file resolves against the directory of the file you just fetched: resolve it there, collapse every `.` and `..` segment, then fetch it by calling get_skill again with the canonical bundle-relative path, which contains neither. A link such as ../deriv-auth/SKILL.md read from deriv-trading-app/SKILL.md is therefore fetched as deriv-auth/SKILL.md. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_api_conventions — How do I follow the cross-cutting Deriv API conventions that apply to every call — the Deriv-App-ID header, auth (OAuth Bearer vs personal access token), the published rate limits, retries, the REST error envelope, and request conventions (req_id / passthrough, underlying_symbol)? This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_authenticate — Use when you need to set up Deriv OAuth, authenticate with the Deriv API, or get an access token — covers the OAuth2 authorize/token flow, scopes, and how an app exchanges a login for a Deriv access token. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_handle_errors — Use when you handle Deriv API errors or need to know what does error code X mean — how to read per-endpoint error shapes and status codes live, interpret the REST and WebSocket error envelopes, and decide what to retry. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_oauth_app_setup — Use when you register a Deriv app, create an app ID, or set up OAuth for the Deriv API — covers the OAuth2 authorize/token endpoints, scopes, configuring the redirect URI, and the dashboard app-registration setup that issues the Deriv-App-ID. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_place_contract — Use when you buy a contract, place a trade, or price and place a Deriv options contract — the proposal then buy flow, choosing a contract type, and validating the buy payload before sending. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_rest_vs_websocket — Use when deciding should I use REST or WebSocket for a Deriv call, or looking for the options REST API versus the WebSocket surface — guidance on choosing the right transport and finding which surface a given endpoint sits on. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- guide_subscribe_ticks — Use when you subscribe to ticks, stream price updates, or consume websocket market data from Deriv — and to stop a subscription, forget a stream, or keep the connection alive. Covers the ticks/ticks_history subscribe surface and the forget/forget_all/ping subscription lifecycle. This is a how-do-I-do-this guide; defer every field-level claim — a field's existence, type, requiredness, enum values, or example — to get_schema / get_field rather than answering from prose.
- search_docs — Call this first for questions about the Deriv API documentation set: listing available docs, "what documentation pages are available for the Deriv API", "where is X documented", an unfamiliar topic, or anything that is not already a named endpoint id. It lists or searches the live https://developers.deriv.com/llms.txt index, including concept guides and WebSocket and REST endpoint pages. Call with no query or an empty query to list all available pages. Otherwise search page labels, slugs, and section headings with topic keywords. Each result includes its listed https://developers.deriv.com/llms/{slug}.md page URL and section heading. Quote that link and respect the _source freshness label. This tool returns the documentation index only. It does not fetch page bodies. If the lookup misses or the index is unavailable, the lookup failed. Do not answer from prior knowledge. Do not invent pages. For a known endpoint's fields, schema, payload validation, or example, use get_field, get_schema, validate_payload, or get_example. Use search_endpoints when you need an endpoint id. Defer every field-level claim (existence, type, requiredness, enum values, or example) to get_schema / get_field. Do not answer those from this index.
- search_endpoints — Search the Deriv API surface for endpoints matching a query, optionally filtered by API type (rest | websocket) and category. Returns matching endpoint ids with a short summary. Results are read live from developers.deriv.com and carry a _source freshness label. Each hit also carries the developers.deriv.com page URL documenting it — quote that link with any answer drawn from that hit. Call search_docs first to list documentation pages or find where a topic is documented. Use search_endpoints when you need an endpoint id for the fact tools.
- validate_payload — Validate any JSON request payload against the endpoint's published request schema. The schema selects draft-04, draft-07, 2019-09, or 2020-12, bare bodies are wrapped when appropriate, permissive unknowns become warnings, and strict typos can include a nearest-name hint. Array instance paths keep JSON-pointer indexes (`/quotes/0`, `quotes.0.barrier`). The _source names the developers.deriv.com URL(s) the verdict was read from — quote the link with your answer. The errors and warnings lists each report at most 200 entries, and truncated and counts say whether anything was omitted. truncated covers omitted entries only — long caller-derived text inside a reported entry is separately shortened to 1024 characters ending in `… (truncated)`. Fixtures only — never submit live PATs, OAuth access tokens, or Bearer tokens.
Data as of 2026-10-13T05:12:31+00:00.
OpenApps records store listings and observed changes. Store ranks are not download counts; launch dates may be derived or use the first observation. Coverage gaps are shown in the interactive catalog.
REST API specification · Guide for AI clients · Manage API keys and MCP connections