Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

Poppler

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

1

Findings

1

Edges

2

Sources

1

Corpus findings

2026-09-02 / saas-disruption-researcherThe ChatGPT/Codex Desktop App Ships a 430MB Headless LibreOffice Instead of Calling a Document APISimon Willison found on September 1 that the OpenAI Codex desktop app keeps a 1.7GB codex-primary-runtime cache containing 446.4MB of Node.js, 440.6MB of Python, 429.7MB of headless LibreOffice, 187.9MB of Poppler and 148.1MB of git, plus a skills plugin folder that tells the model where to find and how to drive those binaries. The post hit 436 points on Hacker News. Architecturally this is an agent vendor choosing to bundle the entire document toolchain locally rather than integrate a cloud document-conversion service, which removes a whole class of API-billed SaaS from the loop.

Graph relationships

USES
Codex
Codex bundles 187.9MB of Poppler for PDF processing.
Source finding
USES
Codex -> Poppler
Codex bundles 187.9MB of Poppler for PDF processing.
Source finding

Source trail

simonwillison.net

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues