IdeaWell

Project ideas from Hacker News discussions.

MCP is dead; long live MCP

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. Centralization & standardization for teams
Organizations need a single, auditable way to expose tools to agents.

“Centralization is Key … we need a standard mechanism of delivering standard capabilities, config, and content across the org.” – jollyllama
“If you have 10‑20 people using agents in wildly different ways, the question of how to baseline the capabilities across your team becomes very real.” – CharlieDigital

2. MCP vs. CLI/skills – the trade‑off debate
Proponents argue MCP gives structured, authenticated calls and telemetry; critics say it bloats context and is unnecessary when a CLI can do the same.

“MCP is fine, particular remote MCP which is the lowest friction way to get access to some hosted service with auth handled for you.” – jswny
“MCP is context bloat and not very good compared to CLIs + skills mechanically.” – jswny (same user)
“MCP is a fixed‑tools protocol… it provides a unified way to connect tools… and forces tools to be explicit about their capabilities.” – kburman

3. Practical value: debugging, telemetry, and agent efficiency
Users report real productivity gains from having a thin MCP wrapper around custom services.

“All the code I work on now has an MCP interface so that the LLM can debug more easily… the amount of time it has saved me is unreal.” – fartfeatures
“The AI can use this to create fixtures, check if things are working… quickly check to see if it is a datastore issue or a query layer issue.” – fartfeatures
“MCP’s structured output schema lets agents plan a single‑step program instead of fumbling through multiple exploratory turns.” – menix

These three themes—centralization for teams, the MCP‑vs‑CLI debate, and the tangible productivity benefits—dominate the discussion.

🚀 Project Ideas

MCPCentral Gateway

Summary

  • A hosted MCP gateway that aggregates internal and third‑party MCP servers, providing single‑sign‑on OAuth, telemetry, and versioned resource/prompt catalogs for enterprise teams.
  • Enables consistent tool discovery and observability across multiple agents and CI pipelines.

Details

Key Value
Target Audience Engineering managers and platform teams in mid‑size to large enterprises using AI agents at scale
Core Feature Central MCP catalog with telemetry, role‑based access, and auto‑generated skill documentation
Tech Stack FastAPI + Starlette, PostgreSQL, Redis, OAuth2, OpenTelemetry
Difficulty Medium
Monetization Revenue-ready: usage‑based SaaS (tiered per‑month fees)

Notes

  • HN commenters repeatedly highlighted the need for “centralized gateway with telemetry” and “standardized docs across orgs” (e.g., CharlieDigital: “We need a standard mechanism of delivering standard capabilities, config, and content across the org.”).
  • Solves the “no sustainable hosting for every team” pain point raised by chatmasta and whattheheckheck.
  • Potential for integration with existing CI/CD pipelines and as a base for open‑source extension.

MCPToolbox Builder

Summary

  • A CLI/SDK that auto‑generates MCP servers from existing command‑line tools, wrapping scripts with the MCP JSON schema and publishing them to a public registry.
  • Turns any local CLI into a first‑class MCP tool without manual protocol coding.

Details

Key Value
Target Audience Individual developers and small teams who already maintain CLI utilities but want AI‑agent compatibility
Core Feature One‑click conversion of local executables to MCP tools, automatic documentation of inputs/outputs, and optional remote deployment
Tech Stack Python + Click, Poetry, Docker for isolated execution, GitHub Actions for CI, FastAPI for remote hosting
Difficulty Low
Monetization Hobby

Notes

  • Directly addresses fartfeatures’ use case: “I built an MCP server for my private VCS tool so the LLM can query issues without UI” and the desire for “few‑line JSON snippet” (e.g., whattheheckheck: “Expose an MCP endpoint… a small json snippet in their mcp.json config”).
  • Community love for “making any tool agent‑ready” is evident in many threads.

MCPResourceHub UI

Summary

  • A web‑based UI that lets non‑technical stakeholders browse, search, and subscribe to MCP resources and prompts, with role‑based visibility and version control.
  • Reduces context bloat by letting agents request only the needed resources/prompts on demand.

Details

Key Value
Target Audience Product managers, designers, and non‑engineer staff who need to discover and configure AI‑assisted workflows
Core Feature Interactive catalog of MCP resources/prompts, badge‑based permissions, real‑time usage stats
Tech Stack React + TypeScript, GraphQL API backed by Neo4j, Auth0 for SSO, Serverless functions for on‑the‑fly rendering
Difficulty Medium
Monetization Revenue-ready: freemium (basic catalog) + paid enterprise add‑ons (custom branding, audit logs)

Notes

  • Commenters emphasized “central discovery” and “visibility” (e.g., jollyllama: “Centralization is key… need to ensure a higher baseline of quality and output across the org”).
  • UI could eliminate the manual “look at a doc” step that moralestapia found confusing.
  • Adds practical utility by letting teams avoid context‑bloat complaints.

MCPSandbox Cloud

Summary

  • A managed sandbox environment where users can safely run MCP servers and CLI wrappers in isolated containers, with built‑in credential vaulting and audit trails.
  • Provides per‑team policy enforcement and cost monitoring for AI‑driven tool usage.

Details

Key Value
Target Audience Security‑conscious enterprises and regulated industries (finance, healthcare) adopting AI agents
Core Feature Containerized MCP execution with automatic secret injection, audit logging, and policy-as-code gates
Tech Stack Kubernetes, HashiCorp Vault, Open Policy Agent, Prometheus + Grafana for monitoring
Difficulty High
Monetization Revenue-ready: subscription per‑node with enterprise SLA

Notes

  • Directly answers tptacek’s security concerns: “MCP doesn’t have a good security story” and the need for “deterministic tooling and auditable interfaces” (e.g., rcarmo: “MCP is being folded into API management”).
  • Aligns with the call for “centralization + security” highlighted throughout the discussion.
  • Offers a clear path to adopt MCP without exposing internal secrets, a frequent blocker for corporate adoption.

Read Later