sempods reference implementation for the JVM — pod server, identity service, hosted MCP, clients
Today your photos belong to a photo app, your messages to a messenger, your notes to a note app, your AI conversations to whoever runs the assistant. You are the tenant; the app is the landlord.
A pod inverts that. It is a data space you host, addressed over HTTP, holding structured linked data. Apps and agents come to your data instead of keeping copies of it, and you decide who may read or write what — and can change your mind without losing anything.
This repository holds the specification of that model and its reference implementation.
1. Resources are HTTP URIs. https://example.org/alice/events/summer-party is both the identifier and the address. Dereference it and you get RDF — JSON-LD by default, other serializations by content negotiation.
2. Every statement lives in exactly one context. A context is a named graph, and it is the permission boundary. Not the resource, not the property — the context. One concept carries the whole access-control model.
3. Permissions are grants on contexts: read, write, manage. Apps and agents obtain them through OAuth 2.1 with PKCE; app identity is a did:web document the server resolves. A grant is durable server-side policy — it never travels inside a token.
4. SPARQL, with the sandbox enforced by the server. A query sees exactly the contexts the caller may read, and writes reach exactly the contexts the caller may write. Client-supplied dataset clauses are not trusted.
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/sempods--sempods-kotlin.html)
Self-hosted MCP server: your codebase as a knowledge graph, your secrets in an encrypted vault — both reachable by your AI assistant, and by nobody else.
aoreshkov/kotlin-lib-mcpMCP server exposing the sources, public API and KDoc of Maven-published Kotlin/Java libraries
replenhq/replenSmarter AI Development Workflows
rwnalds/engramSelf-hosted MCP server + dashboard giving Claude Code, Cursor & Hermes agents shared long-term memory over a git-backed markdown vault. Agent-native, Obsidian-compatible second brain — markdown RAG, n
systempromptio/systemprompt-demoAI Governance Infrastructure — open gateway demo. Self-hosted systemprompt.io gateway in a single Rust binary: policy, audit trails, and cost attribution on every AI request, with the Systemprompt Bri
risa-labs-inc/BossConsoleOpen-source, multi-platform harness for AI agents — a native, multi-threaded operator's console (JVM, not Electron) to run Claude Code, Codex, Gemini or OpenCode with a real browser, terminal, editor,
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.