MCP Front-End Implementation #133

VictorYudin · 2025-04-15T18:31:43Z

Description

This PR implements a Model Context Protocol (MCP) front-end for AgentIQ that allows exposing workflow functions as MCP-compatible tools. The MCP front-end follows the same pattern as the existing FastAPI and Console front-ends, making it consistent with the codebase's architecture.

The implementation includes:

Split MCP code into separate files following the existing pattern
Updated CLI documentation to include the MCP front-end
Created example documentation for using the MCP front-end

Motivation and Context

MCP is an open protocol developed by Anthropic that standardizes how applications provide context to LLMs. Adding an MCP front-end to AgentIQ allows seamless integration with other MCP-compatible applications, enhancing interoperability and expanding the ways developers can utilize AgentIQ workflows.

How Has This Been Tested?

The MCP front-end has been tested with the following workflow:

aiq mcp --config_file examples/simple_rag/configs/milvus_rag_config.yml --tool_names mcp_retriever_tool

This successfully started an MCP server that exposed the mcp_retriever_tool function as an MCP tool.

MCP Front-end Tests

The MCP front-end implementation is thoroughly tested with a comprehensive test suite covering all essential components. All 9 tests pass successfully when running pytest.exe tests/aiq/front_ends/mcp.

The test suite includes:

Configuration validation tests ensuring proper handling of default values, custom settings, and parameter constraints
Plugin functionality tests verifying proper initialization and function filtering
Registration tests confirming the correct async context manager pattern implementation
CLI command registration tests validating proper integration with AgentIQ's command system

By Submitting this PR I confirm:

I am familiar with the Contributing Guidelines.
We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license.
- Any contribution which contains commits that are not Signed-Off will not be accepted.
When the PR is ready for review, new or existing tests cover these changes.
When the PR is ready for review, the documentation is up to date with these changes.

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

copy-pr-bot · 2025-04-15T18:31:47Z

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

AnuradhaKaruppiah · 2025-04-15T18:53:18Z

/ok to test

copy-pr-bot · 2025-04-15T18:53:22Z

/ok to test

@AnuradhaKaruppiah, there was an error processing your request: E1

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/1/

AnuradhaKaruppiah · 2025-04-15T19:36:03Z

/ok to test 17dd2f0

copy-pr-bot · 2025-04-15T19:36:06Z

/ok to test 17dd2f0

@AnuradhaKaruppiah, there was an error processing your request: E2

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/2/

…itial

AnuradhaKaruppiah · 2025-04-15T21:54:26Z

/ok to test

copy-pr-bot · 2025-04-15T21:54:29Z

/ok to test

@AnuradhaKaruppiah, there was an error processing your request: E1

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/1/

AnuradhaKaruppiah · 2025-04-15T21:55:52Z

/ok to test 42fe429

Copilot

Copilot reviewed 12 out of 12 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (1)

tests/aiq/front_ends/mcp/test_mcp_front_end_plugin.py:81

[nitpick] The use of the workflow type value ('test_workflow') as a key in the functions dictionary might be unclear. Consider adding a comment or renaming it to clarify its intended role within the function collection.

assert "test_workflow" in functions

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

src/aiq/front_ends/mcp/mcp_front_end_plugin.py

AnuradhaKaruppiah · 2025-04-16T00:54:23Z

Thanks for the PR @VictorYudin

src/aiq/front_ends/mcp/mcp_front_end_plugin.py

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

src/aiq/front_ends/mcp/mcp_front_end_plugin.py

src/aiq/front_ends/mcp/tool_converter.py

src/aiq/front_ends/mcp/mcp_front_end_plugin.py

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

…itial

AnuradhaKaruppiah · 2025-04-21T15:03:27Z

/ok to test

copy-pr-bot · 2025-04-21T15:03:30Z

/ok to test

@AnuradhaKaruppiah, there was an error processing your request: E1

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/1/

AnuradhaKaruppiah · 2025-04-21T15:04:41Z

/ok to test d7fb27d

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

AnuradhaKaruppiah · 2025-04-21T15:24:37Z

/ok to test f255f1e

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

AnuradhaKaruppiah · 2025-04-21T15:42:04Z

/ok to test 9f376b7

AnuradhaKaruppiah · 2025-04-21T15:48:54Z

/merge

This PR implements a Model Context Protocol (MCP) front-end for AgentIQ that allows exposing workflow functions as MCP-compatible tools. The MCP front-end follows the same pattern as the existing FastAPI and Console front-ends, making it consistent with the codebase's architecture. The implementation includes: - Split MCP code into separate files following the existing pattern - Updated CLI documentation to include the MCP front-end - Created example documentation for using the MCP front-end ## Motivation and Context MCP is an open protocol developed by Anthropic that standardizes how applications provide context to LLMs. Adding an MCP front-end to AgentIQ allows seamless integration with other MCP-compatible applications, enhancing interoperability and expanding the ways developers can utilize AgentIQ workflows. ## How Has This Been Tested? The MCP front-end has been tested with the following workflow: ``` aiq mcp --config_file examples/simple_rag/configs/milvus_rag_config.yml --tool_names mcp_retriever_tool ``` This successfully started an MCP server that exposed the `mcp_retriever_tool` function as an MCP tool. ## MCP Front-end Tests The MCP front-end implementation is thoroughly tested with a comprehensive test suite covering all essential components. All 9 tests pass successfully when running `pytest.exe tests/aiq/front_ends/mcp`. The test suite includes: - Configuration validation tests ensuring proper handling of default values, custom settings, and parameter constraints - Plugin functionality tests verifying proper initialization and function filtering - Registration tests confirming the correct async context manager pattern implementation - CLI command registration tests validating proper integration with AgentIQ's command system ## By Submitting this PR I confirm: - I am familiar with the [Contributing Guidelines](https://github.com/NVIDIA/AgentIQ/blob/develop/docs/source/advanced/contributing.md). - We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license. - Any contribution which contains commits that are not Signed-Off will not be accepted. - When the PR is ready for review, new or existing tests cover these changes. - When the PR is ready for review, the documentation is up to date with these changes. Authors: - Victor Yudin (https://github.com/VictorYudin) - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) Approvers: - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) URL: NVIDIA#133 Signed-off-by: Eric Evans <194135482+ericevans-nv@users.noreply.github.com>

This PR implements a Model Context Protocol (MCP) front-end for AgentIQ that allows exposing workflow functions as MCP-compatible tools. The MCP front-end follows the same pattern as the existing FastAPI and Console front-ends, making it consistent with the codebase's architecture. The implementation includes: - Split MCP code into separate files following the existing pattern - Updated CLI documentation to include the MCP front-end - Created example documentation for using the MCP front-end ## Motivation and Context MCP is an open protocol developed by Anthropic that standardizes how applications provide context to LLMs. Adding an MCP front-end to AgentIQ allows seamless integration with other MCP-compatible applications, enhancing interoperability and expanding the ways developers can utilize AgentIQ workflows. ## How Has This Been Tested? The MCP front-end has been tested with the following workflow: ``` aiq mcp --config_file examples/simple_rag/configs/milvus_rag_config.yml --tool_names mcp_retriever_tool ``` This successfully started an MCP server that exposed the `mcp_retriever_tool` function as an MCP tool. ## MCP Front-end Tests The MCP front-end implementation is thoroughly tested with a comprehensive test suite covering all essential components. All 9 tests pass successfully when running `pytest.exe tests/aiq/front_ends/mcp`. The test suite includes: - Configuration validation tests ensuring proper handling of default values, custom settings, and parameter constraints - Plugin functionality tests verifying proper initialization and function filtering - Registration tests confirming the correct async context manager pattern implementation - CLI command registration tests validating proper integration with AgentIQ's command system ## By Submitting this PR I confirm: - I am familiar with the [Contributing Guidelines](https://github.com/NVIDIA/AgentIQ/blob/develop/docs/source/advanced/contributing.md). - We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license. - Any contribution which contains commits that are not Signed-Off will not be accepted. - When the PR is ready for review, new or existing tests cover these changes. - When the PR is ready for review, the documentation is up to date with these changes. Authors: - Victor Yudin (https://github.com/VictorYudin) - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) Approvers: - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) URL: NVIDIA#133 Signed-off-by: Yuchen Zhang <134643420+yczhang-nv@users.noreply.github.com>

This PR implements a Model Context Protocol (MCP) front-end for AgentIQ that allows exposing workflow functions as MCP-compatible tools. The MCP front-end follows the same pattern as the existing FastAPI and Console front-ends, making it consistent with the codebase's architecture. The implementation includes: - Split MCP code into separate files following the existing pattern - Updated CLI documentation to include the MCP front-end - Created example documentation for using the MCP front-end ## Motivation and Context MCP is an open protocol developed by Anthropic that standardizes how applications provide context to LLMs. Adding an MCP front-end to AgentIQ allows seamless integration with other MCP-compatible applications, enhancing interoperability and expanding the ways developers can utilize AgentIQ workflows. ## How Has This Been Tested? The MCP front-end has been tested with the following workflow: ``` aiq mcp --config_file examples/simple_rag/configs/milvus_rag_config.yml --tool_names mcp_retriever_tool ``` This successfully started an MCP server that exposed the `mcp_retriever_tool` function as an MCP tool. ## MCP Front-end Tests The MCP front-end implementation is thoroughly tested with a comprehensive test suite covering all essential components. All 9 tests pass successfully when running `pytest.exe tests/aiq/front_ends/mcp`. The test suite includes: - Configuration validation tests ensuring proper handling of default values, custom settings, and parameter constraints - Plugin functionality tests verifying proper initialization and function filtering - Registration tests confirming the correct async context manager pattern implementation - CLI command registration tests validating proper integration with AgentIQ's command system ## By Submitting this PR I confirm: - I am familiar with the [Contributing Guidelines](https://github.com/NVIDIA/AgentIQ/blob/develop/docs/source/advanced/contributing.md). - We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license. - Any contribution which contains commits that are not Signed-Off will not be accepted. - When the PR is ready for review, new or existing tests cover these changes. - When the PR is ready for review, the documentation is up to date with these changes. Authors: - Victor Yudin (https://github.com/VictorYudin) - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) Approvers: - Anuradha Karuppiah (https://github.com/AnuradhaKaruppiah) URL: NVIDIA#133

VictorYudin added 3 commits April 15, 2025 11:44

Docs

17dd2f0

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

MCP Front End

4e07eae

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

Tests

7632fd2

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

VictorYudin requested a review from a team as a code owner April 15, 2025 18:31

VictorYudin changed the base branch from develop to main April 15, 2025 18:31

VictorYudin changed the base branch from main to develop April 15, 2025 18:32

VictorYudin changed the title ~~MCP (Modular Communication Protocol) Front-End Implementation~~ MCP Front-End Implementation Apr 15, 2025

mdemoret-nv added feature request New feature or request non-breaking Non-breaking change labels Apr 15, 2025

Merge remote-tracking branch 'upstream/develop' into mcp_front_end_in…

42fe429

…itial

AnuradhaKaruppiah requested a review from Copilot April 15, 2025 21:58

Copilot AI reviewed Apr 15, 2025

View reviewed changes

AnuradhaKaruppiah added 2 commits April 15, 2025 15:17

Move the MCP service readme to the guides section

34b2607

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

Add MCP guide to the index

77bf02c

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

AnuradhaKaruppiah reviewed Apr 16, 2025

View reviewed changes

src/aiq/front_ends/mcp/mcp_front_end_plugin.py Show resolved Hide resolved

src/aiq/front_ends/mcp/mcp_front_end_plugin.py Outdated Show resolved Hide resolved

AnuradhaKaruppiah reviewed Apr 16, 2025

View reviewed changes

src/aiq/front_ends/mcp/mcp_front_end_plugin.py Outdated Show resolved Hide resolved

Tool converter

12acacd

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

VictorYudin force-pushed the mcp_front_end_initial branch from 3bd43ba to 12acacd Compare April 16, 2025 17:42

AnuradhaKaruppiah reviewed Apr 16, 2025

View reviewed changes

src/aiq/front_ends/mcp/mcp_front_end_plugin.py Show resolved Hide resolved

Code review

8224e9a

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

VictorYudin and others added 3 commits April 16, 2025 23:15

removed is_workflow

fa6b348

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

Minor type

0966f90

Signed-off-by: Victor Yudin <vyudin@nvidia.com>

Merge remote-tracking branch 'upstream/develop' into mcp_front_end_in…

d7fb27d

…itial

AnuradhaKaruppiah added 2 commits April 21, 2025 08:18

Fix ci-check errors

3c62f30

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

Fix CI failures in tests

f255f1e

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

More CI fixes

9f376b7

Signed-off-by: Anuradha Karuppiah <anuradhak@nvidia.com>

AnuradhaKaruppiah approved these changes Apr 21, 2025

View reviewed changes

rapids-bot bot merged commit 74f5687 into NVIDIA:develop Apr 21, 2025
10 checks passed

MCP Front-End Implementation #133

MCP Front-End Implementation #133

Uh oh!

Conversation

VictorYudin commented Apr 15, 2025

Description

Motivation and Context

How Has This Been Tested?

MCP Front-end Tests

By Submitting this PR I confirm:

Uh oh!

copy-pr-bot bot commented Apr 15, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 15, 2025

Uh oh!

copy-pr-bot bot commented Apr 15, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 15, 2025

Uh oh!

copy-pr-bot bot commented Apr 15, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 15, 2025

Uh oh!

copy-pr-bot bot commented Apr 15, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 15, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

AnuradhaKaruppiah commented Apr 16, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AnuradhaKaruppiah commented Apr 21, 2025

Uh oh!

copy-pr-bot bot commented Apr 21, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 21, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 21, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 21, 2025

Uh oh!

AnuradhaKaruppiah commented Apr 21, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants