I Tested New Viral Codex Plugin for Claude Code (Shouldn’t Exist But Exploding)
Source: Joe Njenga, I Tested New Viral Codex Plugin for Claude Code (Medium, 2026-04-06)
Key takeaways
- Published before OpenAI’s official Codex plugin launch on 30 March.
- Covers community adoption of Codex-inside-Claude-Code — the viral moment.
- Context for Liu’s deeper, more recent Codex and Claude side-by-side piece.
- Read Liu’s article first for the production-relevant detail; read this one for the “how did this happen” backstory.
Why this is worth knowing but not priority reading
This article documents the community adoption moment for Codex-in-Claude-Code before OpenAI made it official. That context is useful historical colour but Liu’s later article is the technical reference.
If you already read Liu:
- This article adds: early community reception, the “shouldn’t exist but exploding” framing, Njenga’s personal testing narrative
- This article doesn’t add: install commands, cost analysis, adversarial-review examples, Copilot Cowork comparison, SR 11-7 compliance discussion — all of that is in Liu’s piece
The historical moment
- March 2026: Community built informal Codex-in-Claude-Code integrations
- 30 March 2026: OpenAI shipped the official plugin (
openai/codex-plugin-cc) - Njenga’s article: posted 6 April documenting the explosion
- Liu’s article: posted 13 April with production-depth analysis
Read in that order if you want the full arc; skip to Liu if you want the actionable version.
Related Playbook pages
- I Ran Codex and Claude Side by Side — the more thorough deep-read
- Multi-Model Orchestration — the full plugin + orchestration reference