Using LLM Wiki with Codex

Codex should be treated as an execution agent inside a governed repository workflow. It is not the source of truth for the project, and it should not turn raw wiki memory or prior chat history into production changes...

This page is a compact recovered-source summary. The raw source files are retained in sources/recovered/raw/ for audit, while this public page stays small enough for targeted LLM retrieval.

Source And Provenance

• Source domains: llmwikis.org
• Source records in group: 2
• Duplicate resolution: exact-normalized-duplicate
• Primary raw archive: sources/recovered/raw/llmwikis-org/raw-system-archives-llmwikis-source-site-report-preservation-2026-05-01-agent-fi-88272fa84326.md
• Local source paths:
• raw/system-archives/llmwikis/improvement-processing/2026-05-01/staged-site-drafts/using-llm-wiki-with-codex.md
• raw/system-archives/llmwikis/source-site-report-preservation/2026-05-01/agent-file-handoff/Archive/2026-05-01/Improvement/llmwikis-integration-promoted/staged-site-drafts/using-llm-wiki-with-codex.md

Recovered Structure

• Using LLM Wiki with Codex
• Layer Roles
• Recommended Codex Startup
• Promotion Boundary
• What Codex Should Not Do
• Draft Source Basis

Retrieval Notes

• Use this page for orientation and provenance before opening the raw archive copy.
• Treat source-site authority boundaries as active: LLMWikis.org remains the handbook source, UAIX.org remains canonical for UAI-1, and AIWikis.org is the dogfood memory site.
• Unique source variants were not silently discarded; they are listed in the duplicate-resolution report and preserved under the recovered raw archive.

Related Pages

• Recovered Content Index
• UAI Reference Index
• AI Memory Systems Index
• Best Practices Index
• Source Provenance Index