Azure AI Agent ServiceCAUTION
A MCP Server for Azure AI Foundry: it's now moved to cloud, check the new Foundry MCP Server
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
A Model Context Protocol server for Azure AI Foundry, providing a unified set of tools for models, knowledge, evaluation, and more.
[](https://github.com/azure-ai-foundry/mcp-foundry/watchers) [](https://github.com/azure-ai-foundry/mcp-foundry/fork) [](https://github.com/azure-ai-foundry/mcp-foundry/stargazers)
[](https://discord.gg/REmjGvvFpW)
Important notice
MCP Server for Azure AI Foundry (experimental) has moved to the cloud, now as Foundry MCP Server (preview). Please check the official public documentation at Get started with Foundry MCP Server (preview). This GitHub repository contains the outdated information on the previous experimental server implementation, and will not be updated further. Instead, refer to the new, remotely hosted and managed Foundry MCP Server, that has the following benefits:
- Cloud-hosted interface for AI tool orchestration: Foundry MCP Server (preview) provides a secure, scalable endpoint for MCP-compliant clients. You don't need to deploy infrastructure, enabling seamless integration and multi-agent scenarios.
- Identity and access control: The server enforces authentication and authorization with Microsoft Entra ID. It performs all operations within the authenticated user's permissions (On-Behalf-Of flow).
- Scenario-focused, extensible tools: The MCP Server exposes a growing set of tools for read and write operations on models, deployments, evaluations, and
a9161831d0c8OBSERVED · 2026-10-06Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mcp-foundry --env AZURE_OPENAI_API_KEY=${AZURE_OPENAI_API_KEY} -- uvx mcp-foundry{
"mcpServers": {
"mcp-foundry": {
"command": "uvx",
"args": [
"mcp-foundry"
],
"env": {
"AZURE_OPENAI_API_KEY": "${AZURE_OPENAI_API_KEY}"
}
}
}
}Exposed tools (47)
33 read · 11 write · 3 destructive. Blast radius: 3 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
add_document | write | |
agent_query_and_evaluate | read | |
connect_agent | read | |
create_azure_ai_services_account | write | Create an Azure AI services account. |
create_foundry_project | write | Create an Azure AI Foundry Project. |
create_index | write | |
create_indexer | write | |
delete_document | destructive | |
delete_index | destructive | |
delete_indexer | destructive | |
deploy_model_on_ai_services | write | Deploy a model to Azure AI. |
execute_dynamic_swagger_action | write | |
fetch_finetuning_status | read | |
fk_fetch_local_file_contents | read | |
fk_fetch_url_contents | read | |
format_evaluation_report | read | |
get_agent_evaluator_requirements | read | |
get_data_source | read | |
get_document_count | read | |
get_finetuning_job_events | read | |
get_finetuning_metrics | read | |
get_indexer | read | |
get_model_details_and_code_samples | read | |
get_model_quotas | read | Get model quotas for a specific Azure location. |
get_prototyping_instructions_for_github_and_labs | read | |
get_skill_set | write | |
get_text_evaluator_requirements | read | |
list_agent_evaluators | read | |
list_agents | read | List available agents in the Azure AI Agent Service. |
list_azure_ai_foundry_labs_projects | read | |
list_data_sources | read | |
list_deployments_from_azure_ai_services | read | |
list_dynamic_swagger_tools | read | |
list_finetuning_files | read | |
list_finetuning_jobs | read | |
list_index_names | read | |
list_index_schemas | read | |
list_indexers | read | |
list_models_from_model_catalog | read | |
list_skill_sets | read | |
list_text_evaluators | read | |
modify_index | write | |
query_default_agent | read | |
query_index | read | Searches the Azure search index for documents matching the query criteria |
retrieve_index_schema | read | |
run_agent_eval | write | |
update_model_deployment | write | Update an existing model deployment on Azure AI. |
Trust audit
CAUTIONgrade B · trust 89/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (4 observation(s))
- Network
- declared (7 observation(s))
- Shell
- declared (2 observation(s))
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (6)
package = importlib.import_module(base_package)
importlib.import_module(module_name)
ai_search_mcp_server = MCPServerHTTP(url='http://127.0.0.1:8000/sse')
delete_document, delete_index, delete_indexer
> - The server can take `.env` file as an argument to load environment variables from it. You can use the `--envFile` option to specify the path to the `.env` file. If the file is not found, the serve
curl -LsSf https://astral.sh/uv/install.sh | sh
Gates applied: no_behavioural_pass.
a9161831d0c8full audit observations/trust-audit/mcp-server/azure-ai-foundry__azure-ai-agent-service.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-06 | a9161831d0c8 | CAUTION | B | 89 | first audit |
Questions
What is the Azure AI Agent Service MCP server?
A MCP Server for Azure AI Foundry: it's now moved to cloud, check the new Foundry MCP Server
What tools does Azure AI Agent Service expose?
47 in total: 33 read-only, 11 that write, and 3 that can delete or overwrite (delete_document, delete_index, delete_indexer). Every one is listed on this page with its risk.
Is Azure AI Agent Service safe to connect to an agent?
With care. The audit graded it B (89/100) and found 6 things worth knowing before you trust this server, listed below with the exact line each was found on. Separately from the audit: 3 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Azure AI Agent Service need?
It reads AZURE_OPENAI_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does Azure AI Agent Service run?
It speaks streamable-http, so it runs as a service you connect to over the network. It is published on PyPI as mcp-foundry.
How current is this page?
The grade is for one exact copy of the source (a9161831d0c8), read on 2026-10-06. The repository is watched and re-audited when it changes.