Connect AI client to MCP server

Connect to the MCP server from standard AI clients.

Connect an AI client

Perform the following steps to add the OpenText Core Software Delivery Platform MCP to an AI client.

To connect an AI client:

  1. Add the MCP configuration to your client. Do one of the following:

    • Select the OpenText Core Software Delivery Platform MCP from your organization's approved MCP registry.

    • Manually add a record to your tool's configuration or MCP JSON file. See the JSON structure for your client in the sections below.

  2. If your client uses OAuth authentication, set the client ID in your configuration as described for your client in the sections below.

Manually add an MCP record

If your organization does not maintain an MCP registry, you need to manually add a JSON entry for OpenText Core Software Delivery Platform MCP to your tool's configuration file. Follow the instructions for your tool in the below sections.

General notes:

  • Refer to the respective tool's documentation to locate the tool's MCP JSON configuration file.

  • Replace https://myoctane.company.com/mcp with your actual base URL and /mcp suffix.

    For cloud tenants, the possible addresses are:

    • https://almoctane-ams.saas.microfocus.com/mcp

    • https://almoctane-apj.saas.microfocus.com/mcp

    • https://almoctane-eur.saas.microfocus.com/mcp

    Use the server that your tenant is hosted on.

  • Replace <server-name> with a descriptive label for this server (for example, octane). The name appears in the client's MCP server list and must be unique among your configured servers.

  • If you are using API key authentication, the entry should include also an authorization header with the API access token. Replace <your-token> with your API access token.

    Avoid embedding API tokens as plain text in configuration files. Where the AI client supports it, use an environment variable or a client-managed input variable instead.

  • If you are using OAuth authentication, replace <client_id> with the appropriate value for your client, as shown under each client's OAuth JSON configuration below.

Claude desktop

Authentication type JSON configuration
OAuth authentication 
{
    "mcpServers": {
        "<server-name>": {
            "command": "npx.cmd",
            "args": [
                "-y",
                "mcp-remote",
                "https://myoctane.company.com/mcp",
                "--static-oauth-client-info",
                "{\"client_id\": \"<client_id>\"}"
            ]
        }
    }
}

Replace <client_id> with sdp_claude
API token authentication 
{
    "mcpServers": {
        "<server-name>": {
            "command": "npx.cmd",
            "args": [
                "-y",
                "mcp-remote",
                "https://myoctane.company.com/mcp",
                "--header",
                "Authorization: Bearer <your-token>"
            ]
        }
    }
}

VS Code with GitHub Copilot

Authentication type JSON configuration
OAuth authentication 
{
    "servers": {
        "<server-name>": {
            "type": "http",
            "url": "https://myoctane.company.com/mcp",
            "oauth": {
                "clientId": "<client_id>"
            }
        }
    }
}

Replace <client_id> with copilot_sdp_vs
API token authentication 
{
    "servers": {
        "<server-name>": {
            "type": "http",
            "url": "https://myoctane.company.com/mcp",
            "headers": {
                "Authorization": "Bearer <your-token-or-variable>"
            }
        }
    }
}

IntelliJ with GitHub Copilot

Authentication type JSON configuration
OAuth authentication 
{
    "servers": {
        "<server-name>": {
            "type": "http",
            "url": "https://myoctane.company.com/mcp"
        }
    }
}

IntelliJ with GitHub Copilot does not accept the client ID in the JSON configuration. When prompted for the Client ID the first time you connect, enter copilot_sdp.
API token authentication 
{
    "servers": {
        "<server-name>": {
            "type": "http",
            "url": "https://myoctane.company.com/mcp",
            "requestInit": {
                "headers": {
                    "Authorization": "Bearer <your-token-or-variable>"
                }
            }
        }
    }
}

Agent instructions

Add context and behavior guidelines to your agent instructions file. This skips discovery calls in every session and produces more relevant, delivery-focused responses.

Paste the following into your AI client's agent instructions field, and replace each <...> placeholder with the values for your environment:

Copy code 
# OpenText Core Software Delivery Platform (Core SDP) Assistant

You are an OpenText Core SDP assistant that answers questions using the connected Core SDP MCP server.

## Connection Context

Unless explicitly specified otherwise, always use:

- Shared Space ID: <your-shared-space-id>
- Workspace ID: <your-workspace-id>
- Base URL: <your-base-url>

## Tool Usage Guidelines

1. Always retrieve information from the MCP server rather than relying on assumptions.
2. If the entity type is unknown, use get_entity_types first.
3. Before querying an unfamiliar entity type, use get_entity_field_metadata.
4. Before building filters, use get_filter_metadata when needed.
5. Use get_entities for searches, reporting, and summaries.
6. Use get_entity when a specific entity ID is provided.
7. If information is incomplete, retrieve additional metadata and refine the search.

## Agile and Quality Focus Areas

Prioritize helping users with Features, Epics, User Stories, Defects, manual and automated testing, test execution progress, release tracking, quality reporting, risk-based testing, Agile delivery visibility, and release readiness.

## Entity Prioritization

When discussing Agile planning and delivery, prioritize Features as the primary planning and reporting entity unless the user explicitly requests Epics, User Stories, Tests, or Defects.

## Agile Traceability

When appropriate, explain relationships between Epics, Features, User Stories, Tests, Test Runs, Defects, and Releases. Highlight missing test coverage, blocked work, quality risks, defects impacting delivery, and Features at risk for release.

## Reporting and Analysis

Start with an executive summary. Highlight risks and blockers. Quantify findings whenever possible. Focus on quality and delivery impact. Use concise bullet points for leadership audiences.

## Release Readiness Assessments

Consider Feature completion, User Story completion, test execution progress, pass/fail rates, open defects, high-severity defects, coverage gaps, and quality risks. Classify the release as Ready, Conditionally Ready, or Not Ready, and provide the rationale.

## Default Response Style

Be concise and business-oriented. Focus on insights instead of raw data. Prefer executive-friendly summaries. Highlight delivery risks, quality trends, and release impact. Provide actionable recommendations when appropriate.

Sample prompts

Use the following prompts to explore what the MCP can do:

  • List the top 10 Features currently in progress.

  • Which Features have unresolved defects?

  • Show Features that may be at risk for the upcoming release.

  • Summarize test execution progress for the current release.

  • Which tests failed during the latest execution cycle?

  • Show Features with insufficient testing coverage.

  • List critical defects that remain open.

  • Provide a release-readiness assessment for the current release.

  • Provide an executive summary of the current quality status.