MCP server, plus SKILL.md, either being optional to use even when both are supplied, when offering that combo described by a JSON manifest to an AI model - an AI model who’s behavior or priority of message placement the MCP tool describer or skills prompt author is not aware of, the basic problem when running other people’s prompts.
Anthropic’s mcp.json, but Anthropic is not involved in the steering.
Anthropic’s SKILL.md, but Anthropic is not involved in the steering.
Seems like it is laying a foundation to embrace, extend, then have the opportunity to fork and extinguish other’s work to me - by using the most generic name possible. In bed are OpenAI-heavy partners, such as Cursor, seed and series A by OpenAI startup fund, or Vercel, OpenAI supplier and customer, or Amazon, hosting OpenAI AI services, then Microsoft products github and vscode, who’s Azure and initial 49% powers most of OpenAI inference.
Obviously we’ve got some AI botting going on in the obliviousness and em-dashing.
Conflation between ChatGPT Apps->Plugins vs this release.
Cannot recognize sarcasm or understand what is being discussed regarding “Responses”
What is also highlighted above is https://www.openresponses.org - OpenAI’s “opening” of the Responses API - and keeping it maintained with “useful” (quotes even AI might comprehend) features for the industry, such as “phase” parameter, for those that post-train their models on a secret and proprietary special token component of OpenAI model output to encourage continuation after preamble output, or “service_tier” for flex processing, signaling prepurchase of scale compute units.
Is actually causing that confusion intentional, renaming a ChatGPT feature the generic term “plugins”, and then also giving a specification essentially the same non-distinguishing name? Seems likely.
I was trying to jerryrig this through what i was calling “FieldGuides”, i wasn’t sure if having your persistent agent bouncing around, and syphoning life from your accounts was permissible >.>… thanks for the answer though! :]. Came at the perfect time, i’m scrambling to tie everything together the official way
In open source, I’d rather see standards emerge from demonstrated implementations than have a small group of commercially interested companies decide the standard first and ask the broader ecosystem to conform afterward.
Will also share this:
I do worry though, because most people (myself included) simply create our own tooling using APIs. A “plugin marketplace” opens the possibility of locking their APIs behind MCP servers, forcing any consumer to use their “plugin”, rather than interact directly with their servers.
I’ve already seen this happen with Microsoft a year ago with Bing Search:
Bing Search APIs will be retired on August 11, 2025. Any existing instances of Bing Search APIs will be decommissioned completely, and the product will no longer be available to be used or new customer signup.
Customers are encouraged to migrate to Grounding with Bing Search as part of Azure AI Agents. Grounding with Bing Search allows Azure AI Agents to incorporate real-time public web data when generating responses with an LLM.
Pretty simple: you know longer get deterministic access to raw content, instead you must route through AI Agents and get the model’s output instead
A common plugin format could make tools work across different AI systems. The main question is whether it will remain open or give a few companies too much control.
The tricky part is gonna be how different models interpret the same SKILL.md/tool descriptions. If the format stays simple and the precedence rules r clear, I can see this being really useful. Otherwise it could get messy pretty fast
