Pains

Docker cannot automatically manage or enforce the order in which containers launch. While Docker Swarm exists as an orchestration tool, its functionality is limited compared to Kubernetes and only works with Docker containers.

Docker does not provide complete isolation of resources between containers, allowing potential resource interference and conflicts. Additionally, all users with access to the image repository can modify images, creating potential chaos and security issues.

Every tool exposed via MCP requires writing and maintaining a dedicated MCP Server wrapper in Python or TypeScript, plus hosting, updating, securing, monitoring, and scaling. This per-tool overhead accumulates significantly for teams integrating multiple tools.

Installing MCP servers requires finding servers, copying JSON configuration blobs, and manually hard-coding API keys, creating a Byzantine process that serves as a barrier to adoption.

MCP servers can flood conversations with excessive information from searches and operations, quickly exhausting token limits and making conversations unwieldy.

Most API documentation is written for human developers and lacks semantic descriptions needed for AI agents to understand intent. This documentation-understanding gap makes it difficult for LLMs to correctly interpret and use APIs.

Many organizations rely on manual processes at various pipeline stages including code reviews, deployment approvals, and testing. These automation gaps create significant delays in code integration and deployment, rendering CI/CD frameworks ineffective.

Teams are rushing to convert existing REST APIs to MCP servers without considering security implications or efficiency costs. This creates both architectural overhead and expanded attack surface compared to direct API integration.

AI agents lack contextual understanding needed for long-form content and domain-specific nuance, reducing their effectiveness in handling complex scenarios that require deep understanding of broader context.

Live application state often diverges from code definitions (code drift). AI agents struggle to detect and mitigate this without anchoring to reference templates and commit diffs, leading to agents making changes based on outdated or inaccurate code state.

40% of enterprise respondents cited cost of implementation or running costs as a barrier to MCP adoption, making it a significant financial consideration for organizations evaluating the technology.

89% of developers use generative AI daily, but only 24% design APIs with AI agents in mind. APIs are still optimized for human consumers, causing a widening gap as agent adoption outpaces API modernization.

If MCP-like variations proliferate instead of universal adoption, the ecosystem will fragment and vendors will cut corners on compliance. Interoperability breaks down and erodes the security assurances that standardization provides.

Ensuring low-latency, high-throughput communication between distributed MCP components requires sophisticated engineering skills. Performance optimization is a significant barrier for most teams.

When LLMs automatically make API decisions, developers need comprehensive logging and review capabilities for trust and auditing. The lack of transparency into LLM reasoning and generated operations is a critical gap.

MCP servers enumerate all tools and descriptions on first contact, consuming significant tokens during agent discovery. This makes it costly for agents to learn what tools are available compared to human developers.

FastMCP is heavily dependent on the Prefect ecosystem, limiting options for developers who prefer self-hosted or alternative hosting solutions. This tight integration restricts deployment flexibility.

As FastMCP's surface area grows, adding incremental features takes progressively longer due to concerns about breaking existing features or violating spec compliance. The framework architecture makes it harder to introduce new code paths safely.

Tool schemas with JSON Schema references ($ref, $defs) were not being inlined before being sent to MCP clients, violating spec requirements and causing client incompatibilities.

Strategic documents and knowledge work lack version control, audit trails, or decision history. Multiple file versions scatter across shared drives with unclear naming (FINAL, FINAL_v2, FINAL_ACTUALLY_FINAL), and documents become frozen artifacts with no way to understand how decisions were made.

The MCP spec requires all prompt arguments to be strings, but Python functions generating prompts often need structured data (lists, dicts) for business logic. This forces developers to manually parse JSON strings with json.loads() and handle conversion errors.

Organizations find AI-enhanced CI/CD solutions prohibitively expensive for broad deployment. Teams are uncertain about the actual value AI brings, creating resistance to adoption despite recognition of benefits.

The HTTP transport was configured with a 5-second timeout that was cutting connections short for operations that needed more time to complete.

Every tool call requires a round-trip cycle: LLM calls tool, result flows back through context, LLM reasons, calls next tool. Intermediate results that only feed the next step burn tokens repeatedly, reducing efficiency in multi-step workflows.