IBM MCP Integrations
Connect AI applications to IBM products and related tools through MCP.
- Source repo
- IBM/mcp
- Stars
- ★ 414
- Last updated
- 17d ago
- License
- Apache-2.0
- FA score
- 36/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platform
- Setup effort
- High · needs real infrastructure
- You'll need
- Typical use
- An operations team wants an AI client to run IBM MQ health checks or MQSC commands.
- Not a fit if
- Teams seeking one ready-to-run server
- Teams requiring every listed integration to be production-ready
- Source review
- 36/100 · Major gaps 2 safety controls not found
What does this agent do, and when should you use it?
IBM MCP is a directory of IBM MCP servers, clients, and developer tools for connecting AI agents or applications to IBM products and other services. Its listings span automation, business workflows, data, development, infrastructure, monitoring, networking, research, and security. Individual servers expose concrete operations such as querying data, invoking decision services, managing workflows, inspecting cloud resources, or accessing security APIs. The repository describes MCP as an open protocol for models to interact with local and remote resources through standardized servers. It includes both production-ready and experimental servers, and does not provide one installation procedure or a single deployable package.
The repository catalogs separate MCP servers and developer tools, listing their names, descriptions, and links to usage or installation information. Specific servers can expose MCP tools to a client for tasks such as running MQSC commands, querying watsonx.data lakehouses, managing workflow instances, inspecting IBM Cloud resources, or calling security platform APIs. Deployment depends on the chosen project; README examples include local stdio configurations, remote HTTP services, and containerized services. Users configure the connection and credentials in the relevant server and MCP client, then call the server's tools through that client. The repository does not define a common output format, runtime, or deployment boundary.
- An operations team wants an AI client to run IBM MQ health checks or MQSC commands.
- A data team wants conversational access to watsonx.data lakehouse data, catalogs, or table schemas.
- A business automation team wants an assistant to search tasks, read case data, or manage workflow documents.
- A cloud operator wants to inspect IBM Cloud VPC, Code Engine, or Power Virtual Server resources through MCP.
- A security team wants an AI client to query or execute APIs for platforms such as Guardium or QRadar.
How do you install or deploy this agent?
This repository is a catalog, not a package with one installation command or deployable server. Choose an individual server and follow its project or product documentation to configure the runtime, endpoint, and credentials. README examples include uvx, npx, Docker, VS Code installation links, and IBM-hosted remote services, each for particular integrations. Some entries do not include installation steps in this repository.
How do you use this agent?
Add the chosen server to an MCP-capable client and configure its transport and required credentials according to that server's documentation. The README shows VS Code stdio installation configurations for specific servers, including commands, arguments, and environment variables; those are not universal launch instructions for the catalog. Once connected, call the server's tools through the client to query data, inspect resources, or manage workflows. Consult the selected server's documentation for exact tool names, parameters, and a first invocation.
What are this agent's strengths and limitations?
- Covers IBM integrations across automation, business workflows, data, infrastructure, and security.
- Lists concrete operations including SQL queries, workflow management, resource inspection, and REST API calls.
- Shows examples of local stdio, remote HTTP, and containerized integration patterns.
- Includes integrations for IBM products as well as related tools from vendors such as DataStax and HashiCorp.
- It is a directory of separate projects, with no common installer, configuration format, or support boundary.
- The README says the collection includes both production-ready and experimental servers, so maturity varies.
- Many installation instructions live in linked project documentation; setup effort and runtime needs must be checked individually.
- Some servers require product endpoints, API keys, usernames, or passwords.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| IBM MCP Integrations This agent | 36 · Major gaps | MCP server | ★ 414 | 17d ago | — | — |
| Keinsaas Navigator | 69 · Some gaps | Self-hosted serviceFree + model costs | ★ 1.2k | 2mo ago | TypeScript | OpenAI API · Claude API |
| Kronova Asset Intelligence Platform | 30 · Major gaps | Web appFreemium | ★ 252 | 26d ago | TypeScript | — |
| Eko Agentic Workflow | 43 · Major gaps | Library / SDKFree + model costs | ★ 5k | 7mo ago | TypeScript | OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
The README describes MCP services that connect AI agents and lists access to queues, workflows, databases, cloud infrastructure, and security APIs. Some setup examples contain passwords, tokens, or client secrets, but the supplied material does not explain credential protection, permission scope, confirmation before actions, data flows, or rollback, so these scores remain low. There is not enough material to assess dependency security; no evidence supports a red-line finding.
The README distinguishes production-ready from experimental services, but does not label each entry, and uses varying descriptions such as MCP and Model Control Protocol. These weaken consistency. Links and setup snippets offer limited clues about dependencies and availability, while most entries say only to see an external link. No failure messages or handling guidance are provided.
Categories cover automation, business automation, data, development, infrastructure, monitoring, networking, research, security, and tools, with descriptions of varied use cases. This supports discovery across audiences and scenarios. Most entries do not explain capability or permission boundaries. The collection is framed around MCP services and clients, but supported clients, runtime environments, and version requirements are inconsistent or incomplete.
The README's categories and tables aid browsing, but entries vary in detail and many installation instructions are external or missing. Names and descriptions support basic identification, but there is no shared naming or versioning scheme; examples are mostly a few VS Code install configurations, and there is no FAQ. The README gives limited caveats by identifying experimental services and one early-preview product, but no comprehensive limitations. The repository includes an Apache 2.0 LICENSE, justifying full marks for licensing. There is no changelog; a contribution-guide link and community entry provide a limited maintenance path without naming maintainers or an update process.
The directory collects MCP services and tools across many areas and can help users discover integration options in the IBM and related ecosystem. Descriptions and a few clickable setup configurations help, but most capabilities, output formats, and deployment steps require separate research. The supplied material is insufficient to establish the directory's value relative to its maintenance and integration costs.
Service names, brief capability descriptions, and outbound links make some claims traceable, but the review material contains only the README and LICENSE, with no implementations, tests, or linked project documentation for corroboration. Claims about experimental status, production readiness, and capabilities lack supporting verification in the supplied files, and the README does not consistently distinguish verified facts from summaries, so these scores remain low.
- Not found in source: dependency securityPin versions and run a dependency audit (npm audit, pip-audit) before installing; prefer running it in a container.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- This repository is a multi-project directory; the supplied material does not independently verify the capabilities or production-readiness claims in the README.
- Some install snippets include password, token, or secret placeholders. Before deployment, check each service's required permissions, credential handling, action confirmation, and recovery mechanisms.