Model Context Protocol server for Canonical Knowledge Structure.

cks-mcp is an MCP (Model Context Protocol) server that provides LLMs with structured, verifiable knowledge operations through the CKS ecosystem. It exposes six tools—validate, query, compare, evolve, derive, and construct—each backed by the deterministic, immutable semantics of cks-core.

---

Why cks-mcp?

LLMs generate plausible but unverified statements. cks-mcp gives them a canonical knowledge backbone: every piece of information must be explicitly structured, validated against formal constraints, and traceable to its origin. This minimises hallucinations and makes AI‑ generated knowledge auditable.

---

Installation

pip install cks-mcp

The server requires cks-core (installed automatically as a dependency).

---

Quick Start

Launch the server:

cks-mcp

An MCP client (Claude Desktop, any MCP-compatible LLM) can then connect and call tools.

---

Available Tools

Tool	Description
validate_knowledge	Validate a Knowledge Structure and return diagnostics.
query_relations	Find all relations for a given entity.
compare_structures	Check semantic equivalence of two structures.
evolve_knowledge	Apply Genesis/Decay operators to evolve a structure.
derive_knowledge	Derive a new Knowledge Object from existing premises.
construct_knowledge	Parse and construct a Knowledge Structure (coming soon).

---

Usage Example

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "validate_knowledge",
    "arguments": {
      "json_data": "{\"objects\":[{\"identity\":{\"id\":\"obj-1\",\"type\":\"Definition\",\"name\":\"Test\"},\"structure\":{}}]}"
    }
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": "{\"valid\": true, \"error_count\": 0, \"warning_count\": 0, \"diagnostics\": []}"
}

---

Testing

python -m pytest -v

20 tests, all passing.

---

Ecosystem

• cks-core — the canonical knowledge engine (repo)
• CKS Specifications — formal theory behind the system (DOI)

---

License

MIT