Mcp Zig Reference Documentation
Mcp
Current Version: 11.6.1
Chilkat.Mcp
Read resources with
Pass a connected A client for the Model Context Protocol (MCP).
Chilkat.Mcp connects to an MCP server and gives an application access to the
server's tools, resources, and prompts
over a single, unified API. MCP is an open protocol that standardizes how applications
supply context and callable tools to AI models.
Connect & handshake
Connect opens a Streamable HTTP session and performs
the MCP initialize handshake, negotiating the protocol version and
establishing a session id.
Resources & prompts
ListResources and
ReadResource, and fetch prompts with
ListPrompts and
GetPrompt.
Use with Chilkat.Ai
Chilkat.Mcp to Chilkat.Ai.UseMcp to register the
server's tools with an AI model. During a conversation the model's tool calls are
routed to the MCP server automatically.
Streamable HTTP transport
(protocol version 2025-06-18). A server response may arrive as a single JSON
object or as a Server-Sent-Events stream; both are handled transparently. The stdio
transport is not supported.
AuthToken property before calling
Connect. Chilkat then sends it as an
Authorization: Bearer <token> header on every request. Leave it
unset for servers that do not require authentication.
tools/list, resources/list, and prompts/list
results include a nextCursor string. In this release the list methods
return the first page only and do not accept a cursor argument, so nextCursor
is informational.
Object Creation
// Add the package once:
// zig fetch --save https://chilkatdownload.com/11.6.1/chilkat-zig-11.6.1.tar.gz
// and in build.zig:
// const chilkat = b.dependency("chilkat", .{ .target = target, .optimize = optimize });
// exe.root_module.addImport("chilkat", chilkat.module("chilkat"));
const chilkat = @import("chilkat");
// Once per process, before any other Chilkat call:
try chilkat.unlockBundle("Anything for 30-day trial");
const mcp = try chilkat.Mcp.init();
defer mcp.deinit();Creates the underlying native Chilkat object. Mcp is a one-pointer struct passed by value; copying it copies the handle (two names for one object). Returns error.OutOfMemory if the library could not allocate the object. Use an object from one thread at a time; it may be handed from one thread to another.
Releases the native object. Call it exactly once per object (usually with defer); the handle is invalid afterwards. Objects returned by methods are owned by the caller too and are released the same way.
Wraps a handle obtained from the C API (chilkat.c.CkMcp), taking ownership of it. The struct's handle field goes the other way, for anything the Zig API does not cover.
Errors and memory
Methods that can fail return an error union: a method whose only outcome is success or failure returns Error!void; a method producing a string returns (Error || Allocator.Error)![:0]u8; a method producing an object returns Error!T. chilkat.Error is error{ChilkatFailed}; the reason for a failure is in getLastErrorText, and the object remains usable. Properties never fail, and methods that answer a question (hasMember, isUnlocked, ...) return a plain bool.
String arguments are [:0]const u8 (UTF-8; string literals can be passed as is). String results are copies allocated with the allocator argument and owned by the caller, so they stay valid across later calls on the same object.
const text = mcp.someMethod(allocator, ...) catch |err| {
const why = try mcp.getLastErrorText(allocator);
defer allocator.free(why);
std.debug.print("{s}\n", .{why});
return err;
};
defer allocator.free(text);Properties
AuthToken
pub fn getAuthToken(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
pub fn setAuthToken(self: Mcp, value: [:0]const u8) void
An optional bearer token for authenticating to the MCP server. When set, it is sent as an Authorization: Bearer <token> header on every request. The default is the empty string.
Connected
pub fn getConnected(self: Mcp) bool
ConnectTimeoutMs
pub fn getConnectTimeoutMs(self: Mcp) i32
pub fn setConnectTimeoutMs(self: Mcp, value: i32) void
The maximum amount of time, in milliseconds, to wait to establish the TCP/TLS connection to the MCP server. The default value is 30000 (30 seconds).
DebugLogFilePath
pub fn getDebugLogFilePath(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
pub fn setDebugLogFilePath(self: Mcp, value: [:0]const u8) void
If set to a file path, this property logs the LastErrorText of each Chilkat method or property call to the specified file. This logging helps identify the context and history of Chilkat calls leading up to any crash or hang, aiding in debugging.
Enabling the VerboseLogging property provides more detailed information. This property is mainly used for debugging rare instances where a Chilkat method call causes a hang or crash, which should generally not happen.
Possible causes of hangs include:
- A timeout property set to 0, indicating an infinite timeout.
- A hang occurring within an event callback in the application code.
- An internal bug in the Chilkat code causing the hang.
HeartbeatMs
pub fn getHeartbeatMs(self: Mcp) i32
pub fn setHeartbeatMs(self: Mcp, value: i32) void
The interval in milliseconds between each AbortCheck event callback, which enables an application to abort a method before it completes. By default, HeartbeatMs is 0, meaning no AbortCheck event callbacks fire.
IdleTimeoutMs
pub fn getIdleTimeoutMs(self: Mcp) i32
pub fn setIdleTimeoutMs(self: Mcp, value: i32) void
The maximum amount of time, in milliseconds, to wait for additional incoming data when receiving a response has paused. This is not a total timeout. The default value is 180000 (180 seconds).
LastErrorHtml
pub fn getLastErrorHtml(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Provides HTML-formatted information about the last called method or property. If a method call fails or behaves unexpectedly, check this property for details. Note that information is available regardless of the method call's success.
topLastErrorText
pub fn getLastErrorText(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Provides plain text information about the last called method or property. If a method call fails or behaves unexpectedly, check this property for details. Note that information is available regardless of the method call's success.
LastErrorXml
pub fn getLastErrorXml(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Provides XML-formatted information about the last called method or property. If a method call fails or behaves unexpectedly, check this property for details. Note that information is available regardless of the method call's success.
topLastMethodSuccess
pub fn getLastMethodSuccess(self: Mcp) bool
pub fn setLastMethodSuccess(self: Mcp, value: bool) void
Indicates the success or failure of the most recent method call: true means success, false means failure. This property remains unchanged by property setters or getters. This method is present to address challenges in checking for null or Nothing returns in certain programming languages. Note: This property does not apply to methods that return integer values or to boolean-returning methods where the boolean does not indicate success or failure.
ProtocolVersion
pub fn getProtocolVersion(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
pub fn setProtocolVersion(self: Mcp, value: [:0]const u8) void
The MCP protocol version. May be set before Connect to request a specific version; after Connect it reflects the version negotiated with the server and is sent in the MCP-Protocol-Version header of each request. The default is 2025-06-18.
ServerInstructions
pub fn getServerInstructions(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Read-only. Optional usage instructions provided by the server in the initialize result (may be the empty string), available after Connect.
ServerName
pub fn getServerName(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
ServerVersion
pub fn getServerVersion(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
SessionId
pub fn getSessionId(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Read-only. The Mcp-Session-Id assigned by the server during the Connect handshake, or the empty string if the server is stateless. Chilkat automatically includes this session id on all subsequent requests.
VerboseLogging
pub fn getVerboseLogging(self: Mcp) bool
pub fn setVerboseLogging(self: Mcp, value: bool) void
If set to true, then the contents of LastErrorText (or LastErrorXml, or LastErrorHtml) may contain more verbose information. The default value is false. Verbose logging should only be used for debugging. The potentially large quantity of logged information may adversely affect peformance.
Version
pub fn getVersion(self: Mcp, allocator: Allocator) Allocator.Error![:0]u8
Methods
CallTool
Calls the MCP tools/call method for the tool named name, passing the JSON arguments object in args_json. The members of args_json are the tool's input arguments and should match the tool's inputSchema as returned by ListTools. For example, a weather tool might be called with the arguments { "location": "Boston, MA", "units": "fahrenheit" }. Pass an empty JSON object (not a null) for a tool that takes no arguments.
On success, the tool result is placed in result_json. It contains a content array, an isError boolean, and optionally a structuredContent object. Each element of the content array has a type ("text", "image", "audio", or "resource"); for a "text" element the value is in its text member. A typical result looks like:
{
"content": [ { "type": "text", "text": "The weather in Boston is 72F and sunny." } ],
"isError": false
}
Thus the text of the first content item is at JSON path content[0].text.
Note: A tool that executes but reports a failure returns true here with isError set to true in result_json (the failure message is typically carried in the content array). The return value is false only for transport- or protocol-level failures.
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
Close
Explicitly terminates the MCP session by sending an HTTP DELETE with the Mcp-Session-Id header, then releases the connection. It is safe to call this method when not connected (it simply releases any transport). A server that does not permit client-terminated sessions responds with HTTP 405, which is not treated as an error.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
Connect
Connects to the MCP (Model Context Protocol) server at the Streamable HTTP endpoint URL in url (for example https://example.com/mcp) and performs the MCP initialize handshake.
On success, the SessionId, ProtocolVersion, ServerName, ServerVersion, and ServerInstructions properties are populated, and Connected becomes true.
If an AuthToken is set, it is sent as a bearer token on the request. The URL must begin with http:// or https://; the host, port, and path are taken from it. Only the Streamable HTTP transport is supported; the stdio transport is not.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
GetLastJsonData
Returns, in json, the complete JSON-RPC response envelope from the most recent request (the jsonrpc, id, and result or error members), for diagnostics and logging. Note that this is the full JSON-RPC envelope, whereas ListTools, CallTool, ReadResource, and GetPrompt each return only the inner result object.
GetPrompt
Calls the MCP prompts/get method for the prompt named name, passing the JSON arguments object in args_json (pass an empty object for a prompt that takes no arguments). On success, the result is placed in json, consisting of an optional description and a messages array in which each message has a role and content.
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
ListPrompts
Calls the MCP prompts/list method. On success, the server's result object is placed in json, containing a prompts array in which each prompt has a name, optionally a description, and an arguments array, and possibly a nextCursor for pagination.
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
ListResources
Calls the MCP resources/list method. On success, the server's result object is placed in json, containing a resources array in which each resource has a uri, a name, and optionally a description and mimeType, and possibly a nextCursor for pagination.
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
ListTools
Calls the MCP tools/list method. On success, the server's result object is placed in json, containing a tools array in which each tool has a name, a description, and an inputSchema (a JSON Schema describing the tool's arguments), and possibly a nextCursor for pagination.
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
ReadResource
Calls the MCP resources/read method for the resource identified by the URI in uri. On success, the result is placed in json, containing a contents array whose entries carry a uri, a mimeType, and either text (for text resources) or blob (base64 for binary resources).
Connect must be called first.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
SetConnectionSettings
Copies connection settings, including HTTP proxy, SOCKS proxy, bind (network interface) IP address, and TLS options, from sock. This is optional; if not used, default settings apply. Call this method before Connect.
Returns error.ChilkatFailed on failure; getLastErrorText explains why.
Events
All Chilkat methods are synchronous: the call returns when the work is done. During a call, Mcp raises three events so your program can show progress and offer a way out. Declare any subset of abortCheck, percentDone and progressInfo in a struct of your own and install a pointer to it with setEventHandler:
const Progress = struct {
pub fn percentDone(_: *Progress, pct: i32) bool {
std.debug.print("{d}%\n", .{pct});
return false; // return true to abort the method in progress
}
pub fn progressInfo(_: *Progress, name: [:0]const u8, value: [:0]const u8) void {
std.debug.print("{s}: {s}\n", .{ name, value });
}
};
var progress = Progress{};
mcp.setEventHandler(&progress); // progress must outlive the installation
defer mcp.clearEventHandler();
mcp.setHeartbeatMs(250); // abortCheck 4 times per second during Chilkat callsInstalls handler, a pointer to any struct, as the receiver of this object's events, replacing any handler installed earlier. The dispatch is resolved at compile time: only the methods the struct declares are called, and a struct declaring none of the three is a compile error. The struct must stay alive, at the same address, until clearEventHandler or deinit.
Removes the handler; events are no longer delivered.
AbortCheck fires at regular intervals controlled by the HeartbeatMs property (0, the default, disables it); PercentDone fires when an operation's completion percentage is known; ProgressInfo delivers named progress values. Returning true from abortCheck or percentDone aborts the running method, which then returns error.ChilkatFailed.
Events fire on the thread that called the method, before that method returns. To abort a long operation from another thread, have abortCheck read a std.atomic.Value(bool), or set the object's AbortCurrent property.
AbortCheck
pub fn abortCheck(self: *T) bool
Enables a method call to be aborted by triggering the AbortCheck event at intervals defined by the HeartbeatMs property. If HeartbeatMs is set to its default value of 0, no events will occur. For instance, set HeartbeatMs to 200 to trigger 5 AbortCheck events per second.
Example
const Abort = struct {
stop: std.atomic.Value(bool) = .init(false),
pub fn abortCheck(self: *Abort) bool {
return self.stop.load(.acquire); // another thread may call abort.stop.store(true, .release)
}
};
var abort = Abort{};
mcp.setHeartbeatMs(250); // call abortCheck 4 times per second
mcp.setEventHandler(&abort);
defer mcp.clearEventHandler();PercentDone
pub fn percentDone(self: *T, pct: i32) bool
This provides the percentage completion for any method involving network communications or time-consuming processing, assuming the progress can be measured as a percentage. This event is triggered only when it's possible and logical to express the operation's progress as a percentage. The pctDone argument will range from 1 to 100. For methods that finish quickly, the number of PercentDone callbacks may vary, but the final callback will have pctDone equal to 100. For longer operations, callbacks will not exceed one per percentage point (e.g., 1, 2, 3, ..., 98, 99, 100).
The PercentDone callback also acts as an AbortCheck event. For fast methods where PercentDone fires, an AbortCheck event may not trigger since the PercentDone callback already provides an opportunity to abort. For longer operations, where time between PercentDone callbacks is extended, AbortCheck callbacks enable more responsive operation termination.
To abort the operation, set the abort output argument to true. This will cause the method to terminate and return a failure status or corresponding failure value.
Example
const Progress = struct {
pub fn percentDone(_: *Progress, pct: i32) bool {
// pct ranges from 1 to 100.
std.debug.print("Percent done: {d}\n", .{pct});
return false; // return true to abort the method in progress
}
};
var progress = Progress{};
mcp.setEventHandler(&progress);
defer mcp.clearEventHandler();ProgressInfo
pub fn progressInfo(self: *T, name: [:0]const u8, value: [:0]const u8) void
This event callback provides tag name/value pairs that detail what occurs during a method call. To discover existing tag names, create code to handle the event, emit the pairs, and review them. Most tag names are self-explanatory.
Note: Some Chilkat methods don't fire any ProgressInfo events.
Example
const Info = struct {
pub fn progressInfo(_: *Info, name: [:0]const u8, value: [:0]const u8) void {
std.debug.print("{s}: {s}\n", .{ name, value });
}
};
var info = Info{};
mcp.setEventHandler(&info);
defer mcp.clearEventHandler();