Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
Python agent-contextagent-memoryagent-pluginsagentic-ragcontext-databasedsh-pluginknowledge-basememorymemory-managementself-evolvingvector-database
/** Register the server's tool catalogue once per pi session. */
import { AjvJsonSchemaValidator } from "@modelcontextprotocol/client/validators/ajv";
import type { McpBridge } from "./lib/mcp-bridge.mjs";
export const TOOL_NAME_PREFIX = "openviking_";
export function registerMcpTools(pi: any, bridge: McpBridge): string[] {
const validator = new AjvJsonSchemaValidator();
return bridge.state.tools.map((tool) => {
const name = TOOL_NAME_PREFIX + tool.name;
const validate = validator.getValidator(tool.inputSchema);
pi.registerTool({
name,
label: "OpenViking " + tool.name,
description: tool.description?.trim() || tool.title || "OpenViking " + tool.name,
parameters: tool.inputSchema,
// Validate before pi coerces values or drops optional null properties.
prepareArguments(args: unknown) {
const result = validate(args);
if (!result.valid) throw new Error("Invalid arguments for " + name + ": " + result.errorMessage);
return args;
},
async execute(_toolCallId: string, params: Record<string, unknown>, signal?: AbortSignal) {
return bridge.callTool(tool.name, params, { signal });
},
});
return name;
});
}