# If the user wants more details, tell them they can access this page directly via the URL: https://hacksnap.live/story/49779329

# MCP was always a bad idea?

214 points · 177 comments

[Full discussion](<https://news.ycombinator.com/item?id=49779329>)

[Read original](<https://maharship.com/blog/why-mcp-was-always-a-bad-idea/>)

Category: [Agents & Coding](<https://hacksnap.live/?category=agents-coding>)

## Skept-o-meter & Hotness

Skept\-o\-meter: Low\. Estimated from 10 comments\.

14 comments for the summary\.

Peak rank: \#6

Time in Top 10: 24\.0 hours

Hacksnap ranks recent stories first, then orders each group by points\. Peak rank uses all retained observations\. Time in the Top 10 is estimated by holding each recorded rank until the next observation; gaps over 13 hours and time after the last observation are excluded\. Movement between observations is unknown\.

237 recorded rank observations from 2026\-09\-21T12:01:17\.894498\+00:00 to 2026\-10\-10T23:01:00\.95027\+00:00\.

Hotness — latest 168 recorded Hacksnap ranks:

2026\-09\-30T18:03:34\.83179\+00:00: rank \#87

2026\-09\-30T19:02:02\.291882\+00:00: rank \#87

2026\-09\-30T20:02:49\.392147\+00:00: rank \#88

2026\-09\-30T21:03:42\.990832\+00:00: rank \#86

2026\-09\-30T22:02:11\.294387\+00:00: rank \#87

2026\-09\-30T23:01:52\.722953\+00:00: rank \#87

2026\-10\-01T08:02:24\.797064\+00:00: rank \#87

2026\-10\-01T09:01:35\.899477\+00:00: rank \#90

2026\-10\-01T10:01:12\.485692\+00:00: rank \#90

2026\-10\-01T11:00:47\.508128\+00:00: rank \#90

2026\-10\-01T12:00:42\.477917\+00:00: rank \#89

2026\-10\-01T13:00:51\.124852\+00:00: rank \#89

2026\-10\-01T14:00:59\.01153\+00:00: rank \#89

2026\-10\-01T15:00:51\.048047\+00:00: rank \#90

2026\-10\-01T16:00:53\.034654\+00:00: rank \#91

2026\-10\-01T16:36:37\.180071\+00:00: rank \#92

2026\-10\-01T17:00:23\.382263\+00:00: rank \#92

2026\-10\-01T18:01:44\.046011\+00:00: rank \#93

2026\-10\-01T19:01:47\.420738\+00:00: rank \#94

2026\-10\-01T20:01:39\.585083\+00:00: rank \#93

2026\-10\-01T21:01:45\.872085\+00:00: rank \#94

2026\-10\-01T22:01:37\.214899\+00:00: rank \#93

2026\-10\-01T23:04:11\.395288\+00:00: rank \#96

2026\-10\-02T08:02:26\.819307\+00:00: rank \#97

2026\-10\-02T09:02:51\.51649\+00:00: rank \#99

2026\-10\-02T10:02:12\.346099\+00:00: rank \#100

2026\-10\-02T11:02:22\.794376\+00:00: rank \#100

2026\-10\-02T12:01:14\.681578\+00:00: rank \#100

2026\-10\-02T13:02:29\.511591\+00:00: rank \#100

2026\-10\-02T14:01:40\.936738\+00:00: rank \#99

2026\-10\-02T15:02:30\.225476\+00:00: rank \#99

2026\-10\-02T16:02:47\.060695\+00:00: rank \#98

2026\-10\-02T17:02:31\.351493\+00:00: rank \#100

2026\-10\-02T18:01:22\.271899\+00:00: rank \#101

2026\-10\-02T19:01:23\.681417\+00:00: rank \#102

2026\-10\-02T20:01:33\.801055\+00:00: rank \#104

2026\-10\-02T21:02:55\.453316\+00:00: rank \#107

2026\-10\-02T22:01:50\.233139\+00:00: rank \#108

2026\-10\-02T23:01:20\.268516\+00:00: rank \#106

2026\-10\-03T08:01:30\.186188\+00:00: rank \#107

2026\-10\-03T09:01:57\.91467\+00:00: rank \#107

2026\-10\-03T10:00:55\.732001\+00:00: rank \#106

2026\-10\-03T11:01:22\.538596\+00:00: rank \#106

2026\-10\-03T12:01:08\.789384\+00:00: rank \#107

2026\-10\-03T13:00:35\.273559\+00:00: rank \#107

2026\-10\-03T14:00:48\.425606\+00:00: rank \#107

2026\-10\-03T15:01:04\.472963\+00:00: rank \#107

2026\-10\-03T16:01:56\.56175\+00:00: rank \#108

2026\-10\-03T17:01:36\.651036\+00:00: rank \#105

2026\-10\-03T18:01:58\.24207\+00:00: rank \#105

2026\-10\-03T19:01:19\.196589\+00:00: rank \#105

2026\-10\-03T20:00:57\.313759\+00:00: rank \#104

2026\-10\-03T21:01:04\.44439\+00:00: rank \#103

2026\-10\-03T21:20:01\.585706\+00:00: rank \#103

2026\-10\-03T22:00:15\.503126\+00:00: rank \#103

2026\-10\-03T23:00:55\.584832\+00:00: rank \#103

2026\-10\-04T08:01:00\.327459\+00:00: rank \#103

2026\-10\-04T09:00:59\.321023\+00:00: rank \#104

2026\-10\-04T10:00:42\.975336\+00:00: rank \#104

2026\-10\-04T11:01:03\.76577\+00:00: rank \#104

2026\-10\-04T12:01:45\.307985\+00:00: rank \#104

2026\-10\-04T13:00:49\.176267\+00:00: rank \#104

2026\-10\-04T14:00:56\.952543\+00:00: rank \#104

2026\-10\-04T15:00:54\.739637\+00:00: rank \#105

2026\-10\-04T16:00:51\.403054\+00:00: rank \#106

2026\-10\-04T17:00:43\.166713\+00:00: rank \#106

2026\-10\-04T18:00:46\.315535\+00:00: rank \#105

2026\-10\-04T19:00:54\.768841\+00:00: rank \#105

2026\-10\-04T20:00:51\.070442\+00:00: rank \#106

2026\-10\-04T21:00:28\.550532\+00:00: rank \#105

2026\-10\-04T22:01:36\.217651\+00:00: rank \#107

2026\-10\-04T23:01:09\.400767\+00:00: rank \#107

2026\-10\-05T08:01:51\.387003\+00:00: rank \#107

2026\-10\-05T09:01:40\.528475\+00:00: rank \#107

2026\-10\-05T10:00:56\.290772\+00:00: rank \#107

2026\-10\-05T11:01:05\.919518\+00:00: rank \#107

2026\-10\-05T12:00:56\.590881\+00:00: rank \#107

2026\-10\-05T13:01:05\.197156\+00:00: rank \#109

2026\-10\-05T14:01:04\.085718\+00:00: rank \#109

2026\-10\-05T15:00:50\.814571\+00:00: rank \#109

2026\-10\-05T16:02:09\.54139\+00:00: rank \#108

2026\-10\-05T17:00:35\.336076\+00:00: rank \#108

2026\-10\-05T18:00:20\.445491\+00:00: rank \#108

2026\-10\-05T19:02:43\.604975\+00:00: rank \#109

2026\-10\-05T20:01:09\.777459\+00:00: rank \#109

2026\-10\-05T21:01:47\.650671\+00:00: rank \#110

2026\-10\-05T22:02:59\.93099\+00:00: rank \#111

2026\-10\-06T08:02:11\.210914\+00:00: rank \#111

2026\-10\-06T08:02:14\.354697\+00:00: rank \#111

2026\-10\-06T09:01:21\.688765\+00:00: rank \#113

2026\-10\-06T10:00:58\.047773\+00:00: rank \#113

2026\-10\-06T11:00:41\.689929\+00:00: rank \#114

2026\-10\-06T12:00:41\.7705\+00:00: rank \#115

2026\-10\-06T13:01:24\.509519\+00:00: rank \#115

2026\-10\-06T14:00:26\.950817\+00:00: rank \#115

2026\-10\-06T15:00:49\.066414\+00:00: rank \#116

2026\-10\-06T16:00:58\.166747\+00:00: rank \#118

2026\-10\-06T17:01:17\.009511\+00:00: rank \#118

2026\-10\-06T18:01:43\.791429\+00:00: rank \#119

2026\-10\-06T19:00:46\.048609\+00:00: rank \#119

2026\-10\-06T20:01:14\.687294\+00:00: rank \#120

2026\-10\-06T21:02:13\.414382\+00:00: rank \#120

2026\-10\-06T22:00:40\.926271\+00:00: rank \#119

2026\-10\-06T23:03:26\.867445\+00:00: rank \#121

2026\-10\-07T08:01:41\.339804\+00:00: rank \#122

2026\-10\-07T09:02:38\.026986\+00:00: rank \#124

2026\-10\-07T10:01:35\.069531\+00:00: rank \#124

2026\-10\-07T11:02:07\.770084\+00:00: rank \#124

2026\-10\-07T12:01:14\.996697\+00:00: rank \#123

2026\-10\-07T13:00:47\.45912\+00:00: rank \#122

2026\-10\-07T14:01:18\.307739\+00:00: rank \#122

2026\-10\-07T15:01:57\.541362\+00:00: rank \#125

2026\-10\-07T16:01:28\.474022\+00:00: rank \#125

2026\-10\-07T17:01:15\.036908\+00:00: rank \#126

2026\-10\-07T18:00:53\.300079\+00:00: rank \#126

2026\-10\-07T19:01:49\.456657\+00:00: rank \#127

2026\-10\-07T20:01:59\.032487\+00:00: rank \#127

2026\-10\-07T21:05:36\.655392\+00:00: rank \#129

2026\-10\-07T22:02:27\.105241\+00:00: rank \#130

2026\-10\-07T23:01:17\.452894\+00:00: rank \#130

2026\-10\-08T08:03:55\.083808\+00:00: rank \#130

2026\-10\-08T09:03:58\.68898\+00:00: rank \#129

2026\-10\-08T10:04:05\.364235\+00:00: rank \#130

2026\-10\-08T11:02:04\.410307\+00:00: rank \#129

2026\-10\-08T12:02:36\.39618\+00:00: rank \#129

2026\-10\-08T13:03:09\.206401\+00:00: rank \#129

2026\-10\-08T14:03:30\.113384\+00:00: rank \#130

2026\-10\-08T15:02:54\.444758\+00:00: rank \#128

2026\-10\-08T16:03:13\.067807\+00:00: rank \#127

2026\-10\-08T17:03:47\.92465\+00:00: rank \#126

2026\-10\-08T18:01:50\.360431\+00:00: rank \#126

2026\-10\-08T19:02:55\.719274\+00:00: rank \#127

2026\-10\-08T20:01:36\.359022\+00:00: rank \#127

2026\-10\-08T21:03:34\.000719\+00:00: rank \#129

2026\-10\-08T22:01:13\.459732\+00:00: rank \#129

2026\-10\-08T23:01:25\.318392\+00:00: rank \#128

2026\-10\-09T08:01:53\.314021\+00:00: rank \#129

2026\-10\-09T09:02:59\.93426\+00:00: rank \#130

2026\-10\-09T10:02:22\.652527\+00:00: rank \#129

2026\-10\-09T11:02:04\.807282\+00:00: rank \#129

2026\-10\-09T12:02:16\.769295\+00:00: rank \#129

2026\-10\-09T13:02:30\.35034\+00:00: rank \#131

2026\-10\-09T14:01:34\.215134\+00:00: rank \#131

2026\-10\-09T15:01:27\.564392\+00:00: rank \#131

2026\-10\-09T16:02:06\.561056\+00:00: rank \#132

2026\-10\-09T17:03:06\.673648\+00:00: rank \#134

2026\-10\-09T18:01:02\.851052\+00:00: rank \#134

2026\-10\-09T19:01:11\.250291\+00:00: rank \#135

2026\-10\-09T20:01:32\.91907\+00:00: rank \#135

2026\-10\-09T21:02:08\.376144\+00:00: rank \#137

2026\-10\-09T22:01:19\.253075\+00:00: rank \#139

2026\-10\-09T23:01:31\.990912\+00:00: rank \#139

2026\-10\-10T08:01:00\.593376\+00:00: rank \#138

2026\-10\-10T09:01:45\.527975\+00:00: rank \#140

2026\-10\-10T10:00:53\.799353\+00:00: rank \#140

2026\-10\-10T11:00:35\.65851\+00:00: rank \#139

2026\-10\-10T12:01:11\.169647\+00:00: rank \#139

2026\-10\-10T13:00:54\.008377\+00:00: rank \#139

2026\-10\-10T14:01:43\.208766\+00:00: rank \#141

2026\-10\-10T15:01:24\.825188\+00:00: rank \#141

2026\-10\-10T16:01:10\.213328\+00:00: rank \#140

2026\-10\-10T17:00:48\.84189\+00:00: rank \#139

2026\-10\-10T18:00:57\.55132\+00:00: rank \#139

2026\-10\-10T19:00:52\.379463\+00:00: rank \#138

2026\-10\-10T20:00:21\.993912\+00:00: rank \#138

2026\-10\-10T21:01:10\.657738\+00:00: rank \#139

2026\-10\-10T22:00:50\.433569\+00:00: rank \#138

2026\-10\-10T23:01:00\.95027\+00:00: rank \#139

The article's case that MCP is obsolete is strongest for full terminal agents with broad access, but commenters argue MCP still addresses security, control, auditability, and non\-terminal clients, while direct API/CLI use still carries token, reliability, and credential\-handling costs\.

## The brief

The author argues MCP was a stopgap for weaker LLMs and is now obsolete: better models can write scripts, call APIs directly, and discover CLIs via \-\-help, so most MCP servers should be deleted in favor of direct HTTP APIs/CLIs and content negotiation such as Accept: text/markdown and Accept\-Language for preferred programming language\.

- MCP was released in November 2024 by Anthropic and later donated to the Agentic AI Foundation under the Linux Foundation in 2025\.
- Widespread MCP adoption caused context bloat, leading to workarounds like generic search/execute patterns and credential hubs from Composio, MintMCP, and Pipedream\.
- Improved LLMs can execute code, reason over large codebases, call APIs directly, write scripts, and use \-\-help to discover CLIs, reducing the need for MCP servers\.
- Most remote\-service MCP servers wrap existing APIs; Cloudflare's Code Mode composes MCP calls into sandboxed scripts\.
- The author proposes deleting most MCP servers and standardizing direct HTTP API/CLI use, with agent\-identifying headers and content negotiation for Markdown or language\-specific docs\.
- The article cites Accept: text/markdown adoption and a Vercel engineer's call for Accept\-Language to carry preferred programming language, which Shopify docs now support\.

## Discussion themes

### MCP still matters for constrained or non\-terminal agents

simonw argues the article misses MCP's current value: controlling which services an agent can access, handling authentication without exposing API keys, providing a UI for connecting services, and audit logging\. docheinestages asks what agents without terminal access should do, and honoluluxyz says the author needs solutions for hiding credentials and for agents without shell access\.

Sources: [Comment 49779718](<https://news.ycombinator.com/item?id=49779718>) · [Comment 49783301](<https://news.ycombinator.com/item?id=49783301>) · [Comment 49779503](<https://news.ycombinator.com/item?id=49779503>)

### Direct API/CLI use still has token, cost, and reliability problems

rimeice says frontier models still fumble parameters, burn tokens, and bloat context, and that cost\-effective production models will need extra help for at least six months\. tobyhinloopen describes a wrapper that captures CLI output, truncates it, and lets the LLM query more later to avoid huge outputs\. pjmlp notes SaaS often do not offer CLIs for extension APIs and that restarting processes has performance costs\.

Sources: [Comment 49779541](<https://news.ycombinator.com/item?id=49779541>) · [Comment 49779486](<https://news.ycombinator.com/item?id=49779486>) · [Comment 49779571](<https://news.ycombinator.com/item?id=49779571>)

### MCP can formalize common internal tool interactions and save tokens

rsolva says their company built MCPs that simplify frequent internal tool interactions and save time and tokens\. 0x696C6961 argues it makes no sense to waste tokens having an agent re\-discover and re\-implement its own API client each time\. 0x445442 counters that if the interaction is known, a deterministic program can be written once instead\.

Sources: [Comment 49779780](<https://news.ycombinator.com/item?id=49779780>) · [Comment 49779878](<https://news.ycombinator.com/item?id=49779878>) · [Comment 49779865](<https://news.ycombinator.com/item?id=49779865>)

### MCP's value may be organizational or political, not purely technical

mikeocool says REST APIs with OpenAPI specs, standard auth, and a generic make\-api\-request tool could provide the same capabilities, and that MCP's real value was letting companies say they are doing AI and cutting through internal politics\. simonw agrees he made a similar point, but says he came back to MCP to avoid first solving the problem of sandboxing Bash\.

Sources: [Comment 49782200](<https://news.ycombinator.com/item?id=49782200>) · [Comment 49782604](<https://news.ycombinator.com/item?id=49782604>)

### The proposed header/content\-negotiation approach drew criticism

well\_ackshually objects to using Accept\-Language for programming language preference, calling it an abuse of a standard header and suggesting a dedicated header like X\-Prefers\-Lang instead\. The comment also criticizes Vercel and Shopify for pushing this approach\.

Sources: [Comment 49779552](<https://news.ycombinator.com/item?id=49779552>)

### MCP authentication is also questioned

maharshi365 says MCP auth is not secure either because keys are stored in environment variables or files and remain accessible to the agent\. They suggest a proxy approach where credentials are never stored on the device\.

Sources: [Comment 49779518](<https://news.ycombinator.com/item?id=49779518>)

## Sources & coverage

AI-generated summary · 2026\-09\-21T12:01:05\.402449\+00:00

Based on 14 of 14 usable stored comments, selected by depth and branch activity. This is a sample of the discussion. Article text may also be shortened.

Generated using deepseek\-ai/DeepSeek\-V4\.1\-Flash. Check the linked sources for full context.