The Problem

Modern software increasingly composes independent components at runtime—plugin systems, agent harnesses, microservice meshes—yet lacks formal guarantees about what happens when components are added, removed, or reconfigured. Two failure modes dominate: removing a component leaves side effects behind (temporal composability), and components can't declaratively declare and react to changes in their dependencies (spatial composability).

What This Does

This repository is a preprint paper, not a codebase. It formalizes a programming paradigm for spatiotemporal composability by lifting classical effect and coeffect concepts into runtime mechanisms. The core contributions are revertible effects (every context transformation carries an inverse the runtime tracks) and reactive coeffects (context changes notify components against their coeffect specifications). These unify into a single context type, which extends to a calculus of dynamic composition with metatheory proving composability scales from single components to whole systems.

The paper also describes Cordis, a meta-framework implementing these ideas with a core library for effect tracking and coeffect resolution, plus a declarative component loader with configuration reconciliation and hot module replacement.

How It Is Wired

This is a documentation-only repository. There is no executable code, no internal call graph, no entry points, and no code that touches external systems. The wiring has not been mapped for this repository because there is nothing to map.

The three files are:

  • README.md — the landing page with a link to the paper and a draft-date notice.
  • paper.pdf — the actual preprint, dated August 13, 2026, flagged as under active revision.
  • .gitattributes — Git line-ending and diff configuration.

That is the entire repository. The paper is the deliverable; there is no implementation to run, test, or extend.

How To Use It

There is no setup, configuration, or run step. The repository is a static document. To read the content, open paper.pdf in any PDF viewer. To clone it:

git clone https://github.com/moses-y/paper

The README explicitly warns the content may change substantially and asks readers to cite the latest version before relying on specific results.

Real-World Use

The practical application is for a team designing a plugin architecture or agent framework that needs runtime composition guarantees. The paper's formal calculus gives you a specification to implement against: revertible effects tell you what your runtime must track to roll back a component cleanly, and reactive coeffects tell you how components should declare and receive dependency changes. Cordis, as described, is the reference implementation—useful as a design template even if you build your own.

Code Health & Issues

Measured analysis from static inspection of this repository:

  • Medium/SDLC - No test files detected — untested code paths, repository-wide. Expected for a paper-only repo, but there is no implementation to test.
  • Medium/SDLC - No CI/CD pipeline detected — no automated build/test gate, .github/ or CI config absent.
  • Medium/SDLC - No LICENSE file — unclear usage and redistribution rights for the paper's content. If you plan to reuse or distribute this work, clarify licensing with the author.

Additional observation: the repository is a fork of cordiverse/paper (2,547 stars upstream), but this fork has zero stars and no additional files. The README's "under active revision" notice means any implementation based on this draft should pin to a specific commit.

The Bottom Line

This is a theory paper, not a software project. The formal contributions look substantive for anyone building runtime-composition systems, and the draft is honest about being work in progress. The absence of a license is the main practical blocker for reuse; contact the author before building on it. If you need working code, this repo is the wrong starting point—read the paper, then look for Cordis's actual implementation elsewhere.