Codex Import and Build Workflow
This page defines how Codex should be used to import, verify, and build the LMR website content.
Codex may assist with file creation, link correction, naming consistency, and build setup.
Codex must not reinterpret LMR theory.
Core Rule
Codex is allowed to manage files.
Codex is not allowed to invent theory.
It may organize, rename, link, scan, and report.
It may not introduce new primitives, corridors, operators, mechanisms, dynamics, or interpretations.
Allowed Codex Tasks
Codex may:
- create missing Markdown files
- move files into the correct folders
- normalize PDF asset names
- update relative links
- detect broken links
- generate a file tree report
- compare files against
content/site-map.md - update navigation links
- update downloads links
- check whether referenced PDFs exist
- report missing assets
- report inconsistent status labels
- report accidental duplicate pages
- report inconsistent filename conventions
- report obvious Markdown syntax errors
Prohibited Codex Tasks
Codex must not:
- rewrite LMR definitions
- simplify codex language into standard physics language
- replace structural terms with dynamical terms
- add forces, fields, potentials, particles, mechanisms, or interactions to Tier 1 pages
- invent missing equations
- introduce new corridors
- introduce new operators
- treat diagrams as mechanisms
- treat working notes as codex authority
- move Tier 3 correspondence claims into Tier 1
- turn supplements into governing sources
- make claims of empirical confirmation unless explicitly present in the source document
First Codex Pass
The first Codex pass should be read-only.
Codex should scan the project and report:
- existing file tree
- missing files from
content/site-map.md - files not listed in
content/site-map.md - broken Markdown links
- missing PDF assets
- inconsistent public PDF filenames
- inconsistent relative paths
- duplicate pages
- empty pages
- malformed Markdown tables
- accidental code fences
- inconsistent heading hierarchy
Codex should not modify content during the first pass.
First Codex Prompt
Use this prompt first:
Scan this project as a read-only audit.
Compare the actual file tree against content/site-map.md.
Report:
- files listed in the site map that are missing
- files present but not listed in the site map
- broken Markdown links
- missing PDF assets
- inconsistent public PDF filenames
- duplicate or near-duplicate pages
- malformed Markdown
- accidental code fences
- inconsistent heading hierarchy
Do not rewrite theory. Do not edit files. Do not create new LMR definitions. Do not introduce forces, fields, dynamics, mechanisms, particles, potentials, or interactions into Tier 1 material.
Return a structured report only.
Second Codex Pass
After the read-only audit, Codex may perform mechanical corrections.
Allowed second-pass corrections:
- create missing empty files if requested
- add placeholder pages if requested
- fix broken relative links
- update navigation links
- update site-map entries
- normalize public PDF paths
- update Downloads and Release Registry links
- remove accidental code fences
- correct Markdown heading hierarchy
- convert pasted instruction fragments into clean Markdown where obvious
Codex should report every changed file.
Second Codex Prompt
Use this prompt only after reviewing the first audit:
Apply mechanical site corrections only.
Allowed:
- fix broken Markdown links
- normalize relative paths
- update
content/navigation.md - update
content/site-map.md - update
content/downloads.md - update
content/reference/release-registry.md - remove accidental code fences
- correct Markdown heading hierarchy
- create missing placeholder files only if they are listed in
content/site-map.md
Not allowed:
- rewrite LMR definitions
- change theoretical claims
- introduce new terminology
- add standard physics interpretations
- add forces, fields, dynamics, mechanisms, particles, potentials, or interactions to Tier 1 pages
- move Tier 3 claims into Tier 1
- treat working notes as codex authority
Report every file changed and summarize the reason for each change.
PDF Asset Pass
Once PDFs are exported from Overleaf and placed in public/pdfs/, Codex may run a PDF asset check.
Codex should verify:
- every PDF listed in Downloads exists
- every PDF listed in Release Registry exists
- every PDF path uses underscores
- no public PDF link points to a private Overleaf path
- no public PDF link points to a local machine path
- no public PDF filename contains spaces
- no public PDF filename contains unclear draft labels
PDF Asset Prompt
Use this prompt after PDFs are added:
Scan all PDF references in the project.
Verify that every referenced PDF exists in public/pdfs/.
For public PDF assets, enforce underscores.
Do not rename Markdown pages unless a link is broken.
Do not rewrite theoretical content.
Update only:
- PDF filenames
- PDF links
- Downloads references
- Release Registry references
Report all changed paths.
Theory-Language Audit
Codex may run a language audit to find risky wording.
This pass should report only.
It should not automatically rewrite.
Codex should flag possible Tier 1 violations such as:
- force
- field
- mechanism
- interaction
- particle
- potential
- energy transfer
- propagation
- motion
- attraction
- repulsion
- orbit
- curvature
- spacetime dynamics
- collapse
- measurement causes
These words are not always forbidden, but they require context.
Theory-Language Audit Prompt
Use this prompt as a read-only audit:
Search the site for language that may violate LMR tier discipline.
Flag possible Tier 1 category errors involving:
- forces
- fields
- dynamics
- mechanisms
- interactions
- particles
- potentials
- attraction
- repulsion
- orbits
- propagation
- spacetime curvature
- quantum collapse
- measurement causation
Do not edit files.
For each flagged sentence, report:
- file path
- sentence
- why it may be risky
- suggested safer wording
Do not rewrite the file automatically.
Build Setup Pass
If Codex is used to create the actual website, it should first ask what framework is being used.
Acceptable site frameworks include:
- plain static HTML
- Astro
- Next.js
- VitePress
- Docusaurus
- MkDocs
- Eleventy
The content should remain framework-portable where possible.
Framework Rule
The Markdown content is primary.
The website framework is secondary.
Do not embed theory into framework code where it becomes harder to audit.
Keep LMR content in Markdown files whenever possible.
Final Pre-Launch Audit
Before public launch, Codex should verify:
- all pages build
- all links resolve
- all PDFs download
- no private file paths remain
- no Overleaf links are used as public PDF links
- DOI links are correct where available
- all status labels match the Release Registry
- all paper pages link to Downloads
- all Downloads entries link to the Release Registry
- all Reference pages link to each other correctly
- no page contradicts the Status and Versioning page
- no working note is presented as codex authority
Final Pre-Launch Prompt
Use this before publishing:
Run a final pre-launch audit.
Verify:
- the site builds
- every Markdown link resolves
- every PDF link resolves
- no private paths remain
- no Overleaf temporary links remain
- all status labels match the Release Registry
- Downloads and Release Registry agree
- Navigation and Site Map agree
- Reference pages cross-link correctly
- no Working Note is presented as codex authority
- no Supplement overrides a Paper
- no Tier 3 correspondence claim is presented as Tier 1 ontology
Do not rewrite theory.
Return a launch-readiness report with blocking issues first.
Related Materials
content/site-map.mdcontent/navigation.md- PDF Assets and File Naming
- Release Registry
- Status and Versioning
- Codex Rules