Testing Two MCP Server Schemas: Strict Design Blocks All 10 Bad Calls
11/04/2026 — 11/04, 03:17·1 sources (all)·1 reports (all)
Key takeaways
- A developer wrote a note-taking MCP server in pure Python standard library as two versions with different tool names, descriptions and JSON…
- the strict redesign rejected all ten. The post also flags a misleading metric and argues schema design can be measured quantitatively.
Background & analysis
On November 4, 2026, the AI section of DEV Community published a hands-on experiment in which an author built a notes-style MCP Server twice, both times using nothing but the Python standard library. The two builds differed in how tool names, descriptions and JSON schemas were written, but the server functionality was otherwise the same.
To compare the two designs, the author attacked each one with the ten worst calls a model could plausibly make, then ran an actual model against each design and counted what got through. The loose design rejected none of the ten bad calls, for a rejection rate of 0/10. The strict redesign rejected all ten, for a rate of 10/10.
The article also points to a metric that the author describes as pointing in the wrong direction, and uses it to argue that schema design can be measured quantitatively rather than judged by feel. As of the available material, this remains a single report from one independent author on a community site, with no replication, no follow-up coverage, and no verification by other sources.
AI-generated from 1 reports · updated 2 days ago
- Heat index
- 25
- All sources
- 1
- All reports
- 1
- First seen
- 2 days ago
Reports on this story headlines open the original
The author built the same notes MCP server twice in pure Python stdlib and tested each schema design against ten plausible worst-case model calls. The loose design rejected 0 of 10; the strict redesign rejected all 10, and the write-up flags one metric that pointed the wrong way.
DEV Community · AIAI score 68
Other stories people are talking about
- Heat index 840OpenAI ships Images 2.6 with GIF and SVG output15 sources
- Heat index 675OpenAI says its cyber model pipeline confirmed 1,724 Linux kernel flaws and generated 948 privilege-escalation exploits11 sources
- Heat index 642RisingOpenAI GPT-5.6 Brings Major Agentic and Computer Use Upgrades13 sources
- Heat index 633RisingCat Wu, product lead for Claude Code, joins OpenAI9 sources
- Heat index 630NewSurgeOpenAI releases GPT-6 Cyber cybersecurity model7 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion