GitHub and OIP

What this article explains

This article explains GitHub. It shows how GitHub works with the Object Invocation Protocol (OIP). OIP is a way for any computer program to act. It uses plain Uniform Resource Locators (URLs) and receipts. There is no persistent session. Any program that can open a URL can use OIP.

Plain words

GitHub is a service. It stores software project files. These files include the repository, commits, branches, issues, and pull requests.

The repository (repo) is where operational files live. A commit records a change to these files. A branch is a separate line of work. An issue is a tracked problem or task. A pull request is a proposed change. It is sent for review by others.

How GitHub works with OIP

The miscsubjects.com build uses GitHub. GitHub is one source for operational files. It also stores the change history.

OIP exposes GitHub capabilities as objects. An object is a directory row on miscsubjects.com. These objects allow you to interact with GitHub. You can view file inventory. You can read files. You can edit files. You can commit changes. You can check deployment status.

To use an OIP object, you invoke it. You send a request to a server. A server is a computer program that provides services. The request goes to the /api/dispatch endpoint. An endpoint is a specific URL where an Application Programming Interface (API) can be accessed. An API is a set of rules for software to talk to each other.

You can invoke an object using POST /api/dispatch {key, body}. Or you can use GET /api/dispatch?invoke=KEY&body=.... Each invocation creates a record. This record is stored in an append-only ledger. You get a receipt for each action. You can find your receipt at /api/dispatch?receipt=inv_ID.

OIP and MCP: A Comparison

The Model Context Protocol (MCP) is an open standard. An Artificial Intelligence (AI) model connects to an MCP server. This connection uses a session. The server exposes tools, resources, and prompts. The AI model can call these. MCP is not a content-management system.

OIP differs from MCP. OIP does not use a persistent session. It uses plain URLs and receipts. Any model that can open a URL can act with OIP. This makes OIP very flexible. For GitHub operations, OIP allows direct, stateless interactions. This means each action is independent.

Machine Shape: GitHub Objects in OIP

A GitHub object is machine-native. This means a computer program can easily understand it. It names specific fields. These fields describe the GitHub action.

Here is what a GitHub object looks like:

{
  "repo": "The name of the GitHub repository.",
  "branch": "The specific branch in the repository.",
  "path": "The file path within the repository.",
  "sha": "The SHA hash of the file or commit.",
  "operation": "The action to perform (e.g., 'read', 'write', 'commit').",
  "auth": "Authentication details, often a token. A token is data that proves identity or authorization.",
  "result": "The outcome of the operation.",
  "proof": "Evidence of the operation's completion."
}

For more details on the OIP protocol, see the documentation at /api/articles?slug=oip-protocol.

Latest clarity reviews (live)

Fresh models are sent this article's bundle and asked two separate questions: how clear is the machine JSON, and how clear is the English body. Scores are 0 to 10. The full history is in the append-only ledger.

• 2026-07-03 02:25 · model @cf/meta/llama-3.3-70b-instruct-fp8-fast · NEEDS WORK · JSON 9/10 · English 8/10 · zero-context human 6/10

- gaps named: MCP; OIP protocol details

• 2026-07-02 23:49 · model @cf/meta/llama-3.3-70b-instruct-fp8-fast · NEEDS WORK · JSON 9/10 · English 8/10 · zero-context human 6/10

- gaps named: MCP; OIP protocol details

How the loop self-corrects: a failing review queues a model revision of this article (a new append-only version). A missing concept named by a reviewer queues a brand-new machine-written article, which then enters the same review cycle.