Markovian
БесплатноНе проверенBitcoin-anchored provenance for AI outputs; enables stamping, verifying, and tracing outputs with offline-verifiable canonical roots.
Описание
Bitcoin-anchored provenance for AI outputs; enables stamping, verifying, and tracing outputs with offline-verifiable canonical roots.
README
Bitcoin-anchored, independently verifiable provenance for AI outputs, exposed as MCP tools.
Trust the emitter? No. Prove the emitter. Commit an output to the Markovian chain the moment it is produced, and anyone can later confirm it existed then and has not been altered since, without trusting you, us, or any operator.
A stamp proves provenance, not truth. It proves the data was committed at a time. It does not assert the data is correct.
Two doors, one server
This package is the stdio door. The hosted HTTP door is the same server, with the same tool names, the same parameters and the same returned objects:
https://api.markovianprotocol.com/mcp/
Code written against one works unchanged against the other. Pick a transport, not a semantics. (The trailing slash matters: /mcp returns a redirect that some clients will not follow.)
Tools
markovian_stamp(data, wallet?, label?, derived_from?)— commitdatato the Markovian chain and get back a verifiable, tamper-evident stamp (markovian-provenance/v1). Only the SHA-256 of your data is sent; the raw data is never stored. No wallet or account is needed, the first stamp just works. The returnedmerkle_rootis the handle.markovian_verify(merkle_root)— check a stamp against the public verifier. No key, no account. An unknown or edited root returnsverified: false. Anyone can run it, including a party who does not trust the stamper.markovian_trace(merkle_root)— walk a stamp'sderived_fromlineage. Returns a map of the graph, not a verdict: every node carrieshash_bindsandanchored, every edge carriesedge_verified, andvalidis true only if all of them check out.
Lineage links are bound inside the committed bytes, so an edge cannot be added or removed after the fact without changing the root.
Install
pip install markovian-mcp
Configure (Claude Desktop / Cursor / Claude Code)
Local, over stdio:
{
"mcpServers": {
"markovian": {
"command": "markovian-mcp"
}
}
}
Or point at the hosted server instead, with no install:
{
"mcpServers": {
"markovian": {
"url": "https://api.markovianprotocol.com/mcp/"
}
}
}
Verify without trusting us
Verification is a plain public GET. Nothing in it depends on this package, this server, or our good behaviour:
curl https://api.markovianprotocol.com/verify/<merkle_root>
Change one byte of the stamped data and the root no longer matches. That is the whole guarantee, and it is checkable by someone who thinks we are lying.
More at markovianprotocol.com.
Changelog
0.2.0 — breaking. The tools now speak to the live chain and match the hosted
server exactly. Previously this package shipped local-only hash helpers whose
names collided with the hosted server's but whose signatures and semantics did
not (markovian_stamp(content) returning a locally computed digest, versus
markovian_stamp(data) returning a chain commitment). An agent could not move
between the two. There is now one set of tools with one meaning. The old
canonicalization field also claimed RFC 8785 (JCS) while implementing sorted-key
json.dumps, which is not JCS; that claim is gone rather than restated.
Apache-2.0.
Установка Markovian
У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.
▸ github.com/MarkovianProtocol/markovian-mcpFAQ
Markovian MCP бесплатный?
Да, Markovian MCP бесплатный — установка в пару кликов через Unyly без оплаты.
Нужен ли API-ключ для Markovian?
Нет, Markovian работает без API-ключей и переменных окружения.
Markovian — hosted или self-hosted?
Доступен hosted-вариант: Unyly запускает сервер в облаке, локальная установка не обязательна.
Как установить Markovian в Claude Desktop, Claude Code или Cursor?
Открой Markovian на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.
Похожие MCP
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
автор: modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
автор: xuzexin-hzCompare Markovian with
Не уверен что выбрать?
Найди свой стек за 60 секунд
Автор?
Embed-бейдж для README
Похожее
Все в категории ai
