Files
oai-swift/oAI/Services/ExternalMCPModels.swift
T
rune 25028e3405 Add env-var support and native HTTP transport for External MCP Servers
External MCP Servers previously only spoke stdio (spawn a local
command + args). Adds:
- env vars for stdio servers (merged into the subprocess environment,
  not embedded in the args string), with a masked key-value editor
- a native Streamable HTTP transport (URL + Bearer token + custom
  headers), so HTTP-based MCP servers like Obsidian's Local REST API
  plugin connect directly without needing npx/Node.js as a bridge

Introduces an MCPTransport abstraction (stdio/HTTP) so ExternalMCPClient
stays transport-agnostic — mirrors how Provider.swift already abstracts
AI backends in this codebase.

Also fixes a real crash found via live testing against Obsidian:
convertInputSchema force-unwrapped a tool parameter's `type`, which
isn't required by JSON Schema — Obsidian's plugin was the first real
server to send a parameter without one. Live-verified end to end
(vault search/read/write/edit) before this commit, per the project's
standing rule to hold external-service-dependent changes until they're
actually confirmed working, not just compiling and passing tests.
2026-08-13 16:05:42 +02:00

226 lines
7.2 KiB
Swift

// SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
// Copyright (C) 2026 Rune Olsen
import Foundation
// MARK: - Server Configuration
/// Which wire protocol an `ExternalMCPServer` uses. `.stdio` fields are `command`/`args`/`env`;
/// `.http` fields are `url`/`bearerToken`/`headers`. Kept as one flat struct rather than an enum
/// with associated values — simpler `Codable` and simpler settings-JSON storage, at the cost of
/// each server config carrying some always-unused fields for its transport.
enum MCPTransportKind: String, Codable, Sendable, CaseIterable {
case stdio
case http
}
struct ExternalMCPServer: Codable, Identifiable, Sendable {
var id: UUID
var name: String
var transportKind: MCPTransportKind
var command: String
var args: [String]
var env: [String: String]
var url: String
var bearerToken: String
var headers: [String: String]
var isEnabled: Bool
var timeout: TimeInterval
var createdAt: Date
init(
id: UUID = UUID(),
name: String,
transportKind: MCPTransportKind = .stdio,
command: String = "",
args: [String] = [],
env: [String: String] = [:],
url: String = "",
bearerToken: String = "",
headers: [String: String] = [:],
isEnabled: Bool = true,
timeout: TimeInterval = 30,
createdAt: Date = Date()
) {
self.id = id
self.name = name
self.transportKind = transportKind
self.command = command
self.args = args
self.env = env
self.url = url
self.bearerToken = bearerToken
self.headers = headers
self.isEnabled = isEnabled
self.timeout = timeout
self.createdAt = createdAt
}
enum CodingKeys: String, CodingKey {
case id, name, transportKind, command, args, env, url, bearerToken, headers, isEnabled, timeout, createdAt
}
/// Custom decoding so servers saved before `transportKind`/`env`/`url`/`bearerToken`/`headers`
/// existed (plain stdio-only configs) still decode — those keys default rather than fail.
init(from decoder: Decoder) throws {
let c = try decoder.container(keyedBy: CodingKeys.self)
id = try c.decode(UUID.self, forKey: .id)
name = try c.decode(String.self, forKey: .name)
command = try c.decode(String.self, forKey: .command)
args = try c.decode([String].self, forKey: .args)
isEnabled = try c.decode(Bool.self, forKey: .isEnabled)
timeout = try c.decode(TimeInterval.self, forKey: .timeout)
createdAt = try c.decode(Date.self, forKey: .createdAt)
transportKind = try c.decodeIfPresent(MCPTransportKind.self, forKey: .transportKind) ?? .stdio
env = try c.decodeIfPresent([String: String].self, forKey: .env) ?? [:]
url = try c.decodeIfPresent(String.self, forKey: .url) ?? ""
bearerToken = try c.decodeIfPresent(String.self, forKey: .bearerToken) ?? ""
headers = try c.decodeIfPresent([String: String].self, forKey: .headers) ?? [:]
}
var slug: String { Self.makeSlug(from: name) }
static func makeSlug(from name: String) -> String {
let s = name
.lowercased()
.components(separatedBy: CharacterSet.alphanumerics.inverted)
.filter { !$0.isEmpty }
.joined(separator: "_")
return s.isEmpty ? "ext" : s
}
/// Splits a raw arguments string into tokens, respecting single/double-quoted
/// segments so arguments containing spaces (e.g. `--root "/Users/x/My Documents"`)
/// survive intact instead of being split on every space.
static func parseArguments(_ input: String) -> [String] {
var args: [String] = []
var current = ""
var inSingleQuotes = false
var inDoubleQuotes = false
for char in input {
if char == "'" && !inDoubleQuotes {
inSingleQuotes.toggle()
} else if char == "\"" && !inSingleQuotes {
inDoubleQuotes.toggle()
} else if char.isWhitespace && !inSingleQuotes && !inDoubleQuotes {
if !current.isEmpty {
args.append(current)
current = ""
}
} else {
current.append(char)
}
}
if !current.isEmpty { args.append(current) }
return args
}
static let reservedSlugs: Set<String> = [
"anytype", "paperless", "calendar", "reminders",
"contacts", "location", "maps", "bash", "web", "read", "write",
"list", "search", "edit", "delete", "create", "move", "copy", "spawn"
]
var isSlugReserved: Bool { Self.reservedSlugs.contains(slug) }
}
// MARK: - Client State
enum MCPClientState: Equatable {
case idle
case connecting
case ready
case error(String)
case crashed
case stopped
}
// MARK: - State Delegate (all callbacks on MainActor)
@MainActor
protocol ExternalMCPStateDelegate: AnyObject {
func clientDidBecomeReady(id: UUID, tools: [MCPToolDefinition], server: ExternalMCPServer)
func clientDidChangeState(id: UUID, state: MCPClientState)
}
// MARK: - Client Errors
enum MCPClientError: LocalizedError {
case notConnected
case invalidResponse(String)
case timeout
case processLaunchFailed(String)
case handshakeFailed(String)
case writeFailed
case invalidConfiguration(String)
var errorDescription: String? {
switch self {
case .notConnected: return "MCP server is not connected"
case .invalidResponse(let s): return "Invalid MCP response: \(s)"
case .timeout: return "MCP request timed out"
case .processLaunchFailed(let s): return "Failed to launch MCP server: \(s)"
case .handshakeFailed(let s): return "MCP handshake failed: \(s)"
case .writeFailed: return "Failed to write to MCP server stdin"
case .invalidConfiguration(let s): return "Invalid MCP server configuration: \(s)"
}
}
}
// MARK: - MCP Protocol Types
struct MCPInitializeResult: Decodable {
let protocolVersion: String
let capabilities: MCPCapabilities
let serverInfo: MCPServerInfo?
}
struct MCPCapabilities: Decodable {
let tools: MCPToolsCapability?
struct MCPToolsCapability: Decodable { let listChanged: Bool? }
}
struct MCPServerInfo: Decodable {
let name: String
let version: String?
}
struct MCPToolsListResult: Decodable {
let tools: [MCPToolDefinition]
let nextCursor: String?
}
struct MCPToolDefinition: Decodable {
let name: String
let description: String?
let inputSchema: MCPInputSchema
}
struct MCPInputSchema: Decodable {
let type: String
let properties: [String: MCPPropertySchema]?
let required: [String]?
}
struct MCPPropertySchema: Decodable {
let type: String?
let description: String?
let `enum`: [String]?
let items: MCPItemsSchema?
struct MCPItemsSchema: Decodable { let type: String? }
}
struct MCPToolCallResult: Decodable {
let content: [MCPContent]
let isError: Bool?
}
struct MCPContent: Decodable {
let type: String
let text: String?
let data: String?
let mimeType: String?
let uri: String?
}