Support editing Azure DevOps pipeline tags in MCP server

[cpAdm]

Issue Type

• Tool or feature request for Remote MCP Server
• Feedback on Remote MCP Server
• Bug in Remote MCP Server

Tool(s)

Proposed new capability

Support editing Azure DevOps pipeline tags within the Pipelines toolset.

Possible implementation options:

1. New tool: pipelines_get_tags

   • Retrieve tags for a pipeline.
   • Return the current tag set for a pipeline definition.

2. New tool: pipelines_add_tags

   • Add one or more tags to a pipeline.
   • Preserve existing tags unless explicitly removed elsewhere.

3. New tool: pipelines_remove_tags

   • Remove one or more tags from a pipeline.
   • Support removing only the specified tags without replacing the full set.

4. New tool: pipelines_set_tags

   • Replace the full tag set for a pipeline.
   • Useful for reconciliation and policy-driven automation.

Alternatively:

5. Enhance an existing write tool
   • Add pipeline tag operations to the grouped pipelines write surface.
   • For example, support actions such as add_tags, remove_tags, or set_tags.

Description

The Azure DevOps MCP Server currently provides pipeline-related capabilities such as running pipelines, creating pipelines, renaming pipelines, and updating build stages. However, there does not appear to be support for editing pipeline tags through MCP.

Pipeline tags are useful metadata for organization, filtering, governance, and automation. They are also a practical coordination mechanism for AI agents and automated systems that need to identify which pipelines they own or manage.

Without pipeline tag editing support, agents can discover pipelines but cannot reliably mark them for ownership, lifecycle, classification, or policy purposes through the MCP surface.

Why this capability is needed

Pipeline tags are a lightweight way to encode metadata directly on a pipeline definition. They are useful both for human organization and for machine-driven workflows.

In particular, agent workflows can use pipeline tags to claim responsibility for certain pipelines and coordinate safely with users or with other agents.

Without pipeline tag editing support:

• Agents cannot mark which pipelines they manage.
• Agents cannot remove outdated ownership or lifecycle tags.
• Governance or policy tags must be maintained outside MCP.
• Users must fall back to manual UI actions or direct REST/API scripting.
• Automated workflows lose a simple metadata channel for coordination.

Examples of useful tags include:

• managed-by:copilot
• agent:release-bot
• owner:deployment-agent
• environment:production
• lifecycle:deprecated

Example scenarios

Scenario 1: Agent ownership

An AI agent is responsible for maintaining a subset of deployment pipelines.

The agent should be able to add tags such as:

• managed-by:copilot
• agent:release-bot

This would allow:

• identification of agent-managed pipelines
• safer coordination with human operators
• avoidance of conflicts with other automation systems

Scenario 2: Governance and classification

A platform team wants to classify pipelines with tags such as:

• environment:production
• compliance:pci
• service:payments

An MCP-enabled workflow should be able to apply and update these tags consistently across pipelines.

Scenario 3: Cleanup and reconciliation

An automation workflow discovers pipelines with outdated metadata.

The agent should be able to:

• remove old tags such as deprecated-owner
• replace inconsistent tag sets
• reconcile pipeline metadata to a desired state

Suggested workflow

1. Agent identifies a pipeline by ID or name
2. Agent reads existing pipeline tags
3. Agent adds, removes, or replaces tags
4. Updated tags are returned for confirmation

This would allow MCP-driven automation to manage pipeline metadata in the same workflow where it already manages other pipeline operations.