IdeaWell

Project ideas from Hacker News discussions.

You said no MCP

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. MCP is just another API spec (often compared to OpenAPI/REST)
Many commenters see MCP as essentially an API with built‑in documentation or RPC‑style semantics, questioning why a new protocol is needed.

“MCP is the NIH non‑standard version of OpenAPI.” – otabdeveloper4
“It's just like API-s, but with built in documentation.” – tiborsaas
“API + RPC” – pwython

2. Codemode provides a safer, more composable way to chain tool calls
Supporters argue that Codemode (a sandboxed scripting mode) solves MCP’s composability problem, avoids exposing raw shells, and reduces token overhead.

“Codemode lets you execute scripts in a runtime where your MCP tools are made available as function calls.” – pjm331
“The core idea of Codemode … is fixing MCP's biggest flaw—its lack of composability.” – praveenvijayan
“Codemode is just a slightly enhanced version of bash: more complex, sure, but likely more robust, secure, and efficient.” – magnio

3. Subagents are useful for context management but add complexity
Debate centers on whether spawning sub‑agents helps keep the main context clean or introduces unnecessary overhead and caching issues.

“Subagents … can be a mis‑step … more bloat … more complexity for the agent to understand.” – skohan
“I find it pretty handy … I run any bash tools in a restricted sandbox … my trust boundary … is precisely the harness vs. the agent.” – mijoharas
“Using sub‑agents means that they can't use that cache … it just wastes money/tokens.” – sunaookami

4. Pi’s shift from minimalism to feature bloat concerns long‑time users
Several users lament that Pi, originally praised for its minimal, extensible design, is now integrating MCP, Codemode, and subagents as core features, straying from its original ethos.

“Pi is now owned by a company … tend to explode code with features while chasing product market fit.” – qudat
“One of pi's main selling points was its minimal nature.” – skohan
“This just reads like Pi becoming Armins and Marios opinions on what a good harness is instead of letting us build our own with the features we want and need.” – spence-s

5. MCP shines in enterprise settings for security, auth, telemetry, and team workflows
Advocates point out that MCP’s standardized auth, centralized tool distribution, and observability make it valuable for organizations, even if individual developers prefer CLI.

“MCP has pretty much taken over there [in enterprise integrations].” – rcarmo
“MCP is about enterprise control of the server side; no advantages on the client side at all.” – CharlieDigital
“ mônetizing … telemetry on who is using which skill … Server‑side rendering of skills …” – CharlieDigital (on MCP Prompts)

🚀 Project Ideas

MCP File Transfer Proxy (MFTP)

Summary

  • Provides a standardized, secure way to upload and download files within MCP interactions without polluting the LLM context.
  • Eliminates the need for ad‑hoc S3 links or base64 hacks by handling file transfer out‑of‑band via signed URLs or temporary storage.
  • Core value proposition: seamless file handling for agents while keeping MCP messages lightweight and secure.

Details

Key Value
Target Audience Developers building MCP servers or agents that need to exchange files (images, documents, binaries)
Core Feature Middleware that intercepts file‑tool calls, stores the file in a secure bucket, returns a short-lived URL/handle to the agent, and cleans up after use
Tech Stack Node.js (or Python FastAPI), AWS S3 / MinIO, JWT‑signed URLs, MCP SDK
Difficulty Medium
Monetization Revenue-ready: SaaS tiered pricing (free tier for low volume, paid for higher throughput and private storage)

Notes

  • HN users complained about the lack of built‑in file support: “Anyone found a good file upload solution for MCP? Or is the best practice to use HTTP to upload files outside MCP?” (pwython) and “We just do inline base64 uploads … until ~100MB files” (hobofan).
  • A generic proxy would let MCP servers focus on business logic while offloading file handling, addressing the pain point for both hobbyist and enterprise users.
  • Could be offered as a drop‑in extension for popular harnesses (Pi, Claude Code, etc.) or as a standalone service.

MCP Spec Explorer

Summary

  • An interactive web playground that visualizes the Model Context Protocol spec, lets users experiment with tool/resource/prompt calls, and generates ready‑to‑use client code snippets.
  • Solves the confusion expressed by newcomers who struggle to grasp what MCP actually is.
  • Core value proposition: lowers the barrier to adoption by turning the spec into an experiential learning tool.

Details

Key Value
Target Audience Developers evaluating MCP, educators, and curious HN readers who want a concrete understanding
Core Feature Live MCP server sandbox with editable examples, auto‑generated documentation, and code generation for popular languages (TS, Python, Go)
Tech Stack React, TypeScript, MCP SDK, Monaco Editor, Vite
Difficulty Low
Monetization Hobby

Notes

  • The thread opens with uwagar: “so much MCPs in the FTA yet not a line about what MCP actually is.” and the_mitsuhiko: “If we fail to explain it, then we need to do a better job explaining it :)”.
  • An explorer would directly address this need, providing a hands‑on way to see MCP in action, similar to Swagger UI for OpenAPI.
  • Could spark discussion on HN about MCP vs OpenAPI and serve as a referral resource for future threads.

Secure MCP Sandbox Runtime

Summary

  • A lightweight, configurable sandbox for executing MCP tool chains (codemode) that isolates filesystem, network, and credential access while providing fine‑grained permission policies.
  • Addresses security concerns about exposing shell access and enables safe composability of tools.
  • Core value proposition: gives agents the power of codemode without compromising host security.

Details

Key Value
Target Audience Agent harness developers (Pi, Claude Code, OpenCode) and security‑conscious enterprises
Core Feature Sandbox runtime (QuickJS or WASM) with capability‑based ACLs, logging, and resource limits; can be plugged into any MCP host
Tech Stack Rust (for WASM) or C++ (QuickJS), Cap’n Proto for policy definition, optional integration with SELinux/AppArmor
Difficulty High
Monetization Revenue-ready: per‑seat licensing for enterprise harnesses, open‑source core with paid support/extensions

Notes

  • Commenters highlighted the danger of shell access: “you don't really want to expose OS shell access as that opens up a huge security attack surface” (hobofan, lelanthran) and “MCP does literally nothing to prevent these sorts of exploits” (otabdeveloper4).
  • Andrewingram noted the benefits of codemode: fewer LLM hops, lower token usage, repeatability. A secure sandbox would let harnesses adopt codemode safely.
  • Enterprises that need strict compliance (e.g., financial, healthcare) would welcome a auditable, policy‑driven execution environment.

MCP Enterprise Hub

Summary

  • A centralized platform for managing MCP servers at scale: telemetry dashboards, role‑based access control, dynamic prompt/skill customization per team or user, and SSO integration.
  • Targets the enterprise desire for governance, observability, and lifecycle management of MCP‑based skills.
  • Core value proposition: turns ad‑hoc MCP deployments into a governed, observable service akin to an internal API gateway.

Details

Key Value
Target Audience Mid‑to‑large companies deploying MCP internally or via third‑party providers
Core Feature Control plane that registers MCP servers, collects invocation metrics, enforces OAuth2 scopes, and serves personalized prompts via templating
Tech Stack Go (backend), React (frontend), PostgreSQL, Redis, OAuth2 provider (Keycloak/Dex), OpenTelemetry
Difficulty Medium
Monetization Revenue-ready: subscription based on number of registered MCP servers or monthly active users

Notes

  • CharlieDigital emphasized enterprise needs: “Little to no benefit will return to you … Enterprise will want some sort of ‘request admin privileges’ scheme… telemetry on who is using which skill.”
  • The hub would solve the “skills delivered via file system … are static” problem by making prompts and tools dynamic and version‑controlled.
  • Aligns with the discussion around ACLs, telemetry, and the hopelessness of static skill files in large orgs.

OpenAPI‑to‑MCP Converter

Summary

  • A tool that takes an existing OpenAPI (or Swagger) specification and automatically generates a compliant MCP server, complete with tool definitions, OAuth2 flow handling, and elicitation support.
  • Enables teams to leverage their current REST investments while gaining MCP benefits without manually writing MCP boilerplate.
  • Core value proposition: reduces duplication of effort and accelerates MCP adoption by reusing trusted API descriptions.

Details

Key Value
Target Audience API owners who want to expose their services to LLM agents via MCP, and developers building MCP adapters
Core Feature CLI / web UI that reads OpenAPI JSON/YAML, scaffolds an MCP server (Node.js/Go), injects auth middleware, and optionally adds elicitation endpoints for user‑driven parameters
Tech Stack Node.js (NestJS) or Go (Gin), OpenAPI Parser, JWT/OAuth2 libraries, MCP SDK
Difficulty Medium
Monetization Hobby (open‑source) with optional paid templates or enterprise support

Notes

  • Multiple users argued MCP is redundant: “MCP is the NIH non‑standard version of OpenAPI.” (otabdeveloper4) and “why does OpenAPI need to exist when IDL could have been extended?” (jasomill).

  • A converter would let them keep their OpenAPI as the source of truth while still providing an MCP façade, directly answering the “.」

  • HN thread also mentioned the need for OAuth handling: “The harness would walk you through the Oauth flow and securely store the token.” (mikeocool). The converter would automate this.

Read Later