Use pointers for the protocol's struct types, to future-proof against
them getting large.
Change-Id: I55b204080a41c8140539697ba264e502e912e613
Reviewed-on: https://go-review.googlesource.com/c/tools/+/672575
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Also, make sure that NewBlobResourceContents creates a Blob.
Change-Id: I2f0fe2ac6410d6b155deb2c8643a40bc73e25cd4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/672418
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
To be compatible with Go 1.23, we cannot use the
encoding/json "omitzero" option. Replace with a
custom MarshalJSON method.
Change-Id: Ied615ea04fb20e2459f9ed20ef03c81edbc74961
Reviewed-on: https://go-review.googlesource.com/c/tools/+/672417
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Now that the protocol package is merged into the mcp package, clean up
the higher level 'Content' union types, in favor of types constructors.
Change-Id: I37bedb0ff0a296eef799ba65c37ca7f94ee1dbbe
Reviewed-on: https://go-review.googlesource.com/c/tools/+/672415
Reviewed-by: Jonathan Amsterdam <jba@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
In accordance with the updated design, merge the protocol package into
the mcp package. To do so, rename protocol.Content to WireContent and
protocol.ResourceContents to WireResource.
Also, remove some stray logging in readResource.
Change-Id: I76f61ce8d67dd3ac520ef2e1959ade0a0785a480
Reviewed-on: https://go-review.googlesource.com/c/tools/+/672375
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Design for pagination.
Change-Id: I8257a499b1e4691382b5681b84a52030344ab0e3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671135
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
When we check the root schema during resolution, first make sure
its descendants form a tree.
Change-Id: Icc430e7df463aa84b961c28616be567f98960430
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670776
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Support remote references.
Most of the machinery was in place; we just had to define the
right loader, copy files from the test repo, and download
the meta-schema files from json-schema.org.
Change-Id: Ib588200ef0fa483641b9b53f9904f53b39dc4b82
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670795
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Implementation of resources.
This follows the design sketched in design/design.md, except for
Server.FileResourceHandler, which will be in a forthcoming CL.
As the spec recommends, the server verifies an incoming URI from the
client against its list of resources, to make sure the URI came from one
of them.
Some things that aren't in the design doc:
- I provide an exported function that creates a "resource not found"
jsonrpc2.WireEror error for server authors to return.
A wrinkle: the MCP spec defines the error code for
"resource not found" as -32002 [1]. That code is reserved for use
by implementations of JSON-RPC, not applications. That matters to us
because our JSON-RPC implementation already uses that code for the
"server closing" error. I filed a bug against MCP about it [2] and
chose a different code for the "resource not found" error.
- As a slight convenience, I fill in some fields in the
ReadResourceResult if the handler doesn't set them.
[1] https://modelcontextprotocol.io/specification/2025-03-26/server/resources#error-handling
[2] https://github.com/modelcontextprotocol/modelcontextprotocol/issues/509
Change-Id: I06153da982ea00748a3691ded923448a880cea2a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671363
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Mention that all methods from the spec have the same signature,
and then don't repeat those method declarations.
Modify all method calls in the doc to take a Params struct, even if
the current MCP spec says it is empty.
Change `string` to `[]byte` in some of the listed struct fields.
The current code generator incorrectly generates these as strings.
(I have a forthcoming CL to fix that.)
Also, elaborate the logging design.
Change-Id: I5f323d4d98a4b6045d3cdef770d13b4bfdcb1044
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671358
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Handle resource subscriptions by delegating to the user.
We don't need to use the word "Resource" in the handlers because
the Subscribe and Unsubscribe methods in the spec don't use that
name. If the spec adds another kind of subscription, they will pick
a new name.
The need for two handlers is dictated by backward compatibility.
We must pass the RPC params to the subscription handler, which means
we need different signatures for subscribe and unsubscribe.
Change-Id: I045abaf52dc9c3e96181331728d50cf36cc8ed91
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671895
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
- Change the names of some content-related types to better match
the spec. Most notably, rename Resource to EmbeddedResource,
because the former has a different meaning in the spec.
- Use []byte intead of string where the spec says "base64-encoded data."
- Unexport EmbeddedResource.ToWire.
Change-Id: I9f65efaadadda1b3abc0e58c64a743af1a72852c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671362
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Commit-Queue: Jonathan Amsterdam <jba@google.com>
You can now write "*" for a name in a config to use the original
name.
Also, fix a bug in dealing with initialisms.
Change-Id: I4d5d0acb1200fff6500644457ec34c5026d63f9f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671361
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Add support for roots.
Change-Id: Ia50abc88f0047238272d698f30ce615b1a8fd486
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671360
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
What the spec calls "lists" of features (tools, prompts and so on)
are actually sets. Define a common data structure we can use to
implement all of them, and use it for server tools and prompts.
Change-Id: I257d0776b87a746ff23cd49ebcc7b317f4ea65e1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671359
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Minor changes to the roots part of the design.
Change-Id: I8125aecc5da297594f74d8d7dc5ef3d94dfa8519
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671356
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Fix typos and formatting, reword slightly.
Also, change names in the middleware section.
Change-Id: Ie86e5f633a07675ab188a8aaf70e51a48b0878bb
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671555
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Put discussion of mcp-go up front, and address differences in the
individual sections. I think this leads to easier reading for the user
who is concerned about our divergence in design.
Change-Id: Ifd21f6b9eb570ad4a325b8a8800b7868dac12205
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671556
TryBot-Bypass: Robert Findley <rfindley@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Commit-Queue: Robert Findley <rfindley@google.com>
Per discussion, remove the separate protocol package in favor of a
single mcp package. Update the tools section accordingly, and add more
discussion about the decision of the NewTool API.
Change-Id: Ifb259555ca9d493ce63bc4f71e5f2fa295bdf09c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671496
TryBot-Bypass: Robert Findley <rfindley@google.com>
Commit-Queue: Robert Findley <rfindley@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
After much deliberation, revert back to the symmetrical many-to-one form
of the client APIs. Every time I tried to write the asymmetrical form, I
couldn't justify it. Furthermore, I think the names 'ClientSession' and
'ServerSession' significantly clarify the meaning.
Change-Id: I5c5e1eaf0d5de5ac7c98dab7aad9c0f31c3ff4e1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671177
Auto-Submit: Robert Findley <rfindley@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Design for the completions feature.
Change-Id: Ia793091a10adb9c51472a9d3d4acdce8d6932536
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671376
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Describe our tool design. This is short because most of the interesting
part is in the JSON Schema section.
Change-Id: I017fca3055aef881285280a379e9f235bf95e75d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671355
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Describe how we use JSON Schema.
Change-Id: I488a09854333ba7280401e3bed5f50f83f481cd2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671215
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Describe the differences with mcp-go, and justify our choices.
Change-Id: Ib9b7ed87b8038f0b7cbc0eeffc5d7a7075eeae30
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671017
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Bypass: Jonathan Amsterdam <jba@google.com>
Discuss generated protocol types, and expand on the section on package
layout to make it clearer how they fit in to the SDK.
Change-Id: I3beaa86c2396e0cee41dcc8eac06d93c54cb96df
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671016
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Expose the SSEServerTransport type, so that clients can customize the
handling of sessions and endpoints by writing their own HTTP handler.
This is similar to the behavior of the typescript SDK.
Also, finish documenting transports in the design doc.
Change-Id: Ief5aa8424ba2946a9615e366969a4750f40224a8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/671015
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Auto-Submit: Robert Findley <rfindley@google.com>
Implement the $ref keyword, except for remote references (those
pointing outside the root schema).
Change-Id: Ia5ca88464e62e9c613970d8d9fcd1512ab288934
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670775
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Enhance the generator so that it recognizes initialisms like "Id" and
writes them in all caps, as is idiomatic for Go.
Also, fix the import path of jsonschema in the generated code.
Change-Id: Ib12e9fbd543d360035d5e324f0aefca584ac7c85
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670916
Auto-Submit: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Design for logging.
Change-Id: I787bb51318337bd09f973e04d0e4dbf526e09fb6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670935
Reviewed-by: Robert Findley <rfindley@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Auto-Submit: Jonathan Amsterdam <jba@google.com>
s.every now visits the entire subtree of schemas at s, not only the
immediate children of s.
This re-introduces a tower: now of closures instead of iterators.
We could remove it having everyChild call every recursively instead
of returning, just passing f down all the way unwrapped.
But this is going to complicate the recursive function inside
resolveURIs, which needs to pass parent context down. It would have
the same problem that ast.Inspect has, and we'd have to adopt the same
solution, maintaining a stack explicitly.
I feel it's cleaner to keep the recursion simple at the cost of building
several closures. To do otherwise would be premature optimization.
Change-Id: Ic453cdaac482da8147dafc34b0117d25458d1c9d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670675
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Start writing up a formal design doc for an MCP SDK based on what we've
learned from the prototype. Include discussion of package layout,
jsonrpc2, and transport. Lay out the rest of the design sections,
leaving stub comments describing what they should cover.
Change-Id: Id0e5b408eac8c0acb5ef90a32d0d69473d8e83f0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670536
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Auto-Submit: Robert Findley <rfindley@google.com>
It's too confusing that you send server requests through the Client, and
send client requests through the ClientConnection. Rename
ClientConnection to ServerConnection, so that the types on the
server side of the control diagram start with 'Server'.
Change-Id: Idfd27af464a9296a2341cd23b2075cd9d0eb6b45
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670538
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Robert Findley <rfindley@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Implement enough of the JSON Pointer spec to suffice for JSON Schema.
For details, see the comment at the top of json_pointer.go.
Change-Id: Ic297f1a1a08903b4b3c6666fe57ac1f45c96c0f7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670535
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The instance passed to Resolved.Validate can be a struct, or can
contain structs. A struct is treated like the JSON object that it
marshals to.
Also support pointers to any depth. That is, if an instance is a **int,
then Validate will treat it like an int (or nil). That won't happen if
the instance is unmarshaled into an `any`, but may if it's unmarshaled
into a struct.
Change-Id: I39f9af58028bd779887495754519615cd5dfb6c8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670395
Reviewed-by: Alan Donovan <adonovan@google.com>
Auto-Submit: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Pass the client Transport to NewClient instead of Connect.
Combine the ConnectionOptions with the ClientOptions.
Change-Id: I3acd6cf6d5bf46666742f2a73e2acaef2057ea77
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670215
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
Move examples from package doc to README.md with package decl and
imports to ensure it's compilable.
Move status of the package from package doc to README.md.
Add core concepts of MCP with example code snippets and testing
recommendations.
Change-Id: I5bf5000b922fa3942aba922713b12d0347295811
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670316
Reviewed-by: Sam Thanawalla <samthanawalla@google.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
The generated protocol package is part of the SDK API, so it shouldn't
be internal.
Change-Id: I5306af48019b1599707b939b348eb6ae4f766ca9
Reviewed-on: https://go-review.googlesource.com/c/tools/+/670356
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Assign a base URI to every schema.
See the comment on resolveURIs for details.
Change-Id: I2332d825c486766832503cabdd0bf9d819171d53
Reviewed-on: https://go-review.googlesource.com/c/tools/+/669997
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
The jsonschema package is part of of the API of the mcp package,
so it should not be internal.
Change-Id: I9520797837329fc2cda09c2f46ec4d48d031b9d3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/669996
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Robert Findley <rfindley@google.com>
This is the first CL that deals with the process of preparing a schema for validation.
Perform some basic checks on the scheme. Along the way, compile regexps and store
them in the schema for use during validation.
Change-Id: I0e5e6ce28656dcecb6d2d4b2fdc98998fa05b6f1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/669696
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alan Donovan <adonovan@google.com>
Determine if a slice contains unique items by hashing them.
This should take linear time on average, better than the previous
quadratic algorithm.
Change-Id: I8dc95bb6f29d802bdbc64bc7f7e698c71eae5ce7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/669455
Reviewed-by: Alan Donovan <adonovan@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
