The Codex desktop app keeps 1.7GB of tools on disk, LibreOffice included
Simon Willison went looking through his cache folder and found that the OpenAI Codex desktop app, since rebranded to ChatGPT, stores a 1.7GB runtime there. Inside it are a full Python installation, a full Node.js installation, and native binaries for Poppler, git and LibreOffice. A neighboring folder holds skills that tell the agent where those binaries are and how to use them. It is a small finding with a large implication about where document work actually happens.
On September 1, 2026, Simon Willison reported that the OpenAI Codex desktop app, since rebranded to ChatGPT, keeps 1.7GB of files in a cache folder named codex-primary-runtime. He found it while looking through his cache with a disk usage tool, which is to say it was never announced. What it holds explains more than its size does.
What is in the folder
The runtime is not a model or a set of weights. It is an ordinary local toolchain, of the sort you would install by hand to process documents on your own machine.
- A full Python installation and a full Node.js installation.
- Native binaries for Poppler, which reads and converts PDF files, and for git.
- The LibreOffice office suite, which opens and converts word processor, spreadsheet and presentation formats.
Why it is there
A folder inside the same runtime, under the documents plugin, contains skills that tell Codex how to find and use those binaries. That is the part that turns a disk observation into a capability one. An agent that ships LibreOffice and Poppler alongside instructions for calling them is being set up to open, convert and write office documents and PDFs locally. It does not need to upload them somewhere and wait for a service to answer.
This suggests two practical consequences. Document conversion in the app does not require the file to leave the machine, which matters for anyone whose documents are not supposed to travel. And the cost is 1.7GB of disk per installation, sitting in a cache folder where nobody thinks to look. Neither is a problem, but both are the kind of thing worth knowing before an agent is rolled out across a fleet of laptops.
„including a full Python installation, a full Node.js installation“
Related
OpenClaw 2026.10.2 stops long tool output from slipping credentials past redaction
A bug in OpenClaw's secret masking let a credential through whenever it straddled the boundary between two slices of a long text. The fix ships in 2026.10.2 along with repaired updates from 2026.10.1 and a new compaction default for OpenAI and Anthropic API keys.
OpenClawverified

OpenClaw 2026.10.1 lets agents search installed skills and changes how sessions message each other
OpenClaw released version 2026.10.1 on October 10, 2026, its first stable release since 2026.9.7. Agents can now search the skills installed on a machine and read only the one a task needs, instead of carrying the whole library in the prompt. Opt-in MCP Apps run beside chat, and full backups can go to an external drive or Cloudflare R2. Three changes can break existing setups: session-to-session messaging, the old required argument, and Workboard data stored before July 2026.
OpenClawverified
Claude Code 2.1.294 fixes prompt hooks that let through the actions they were written to block
Anthropic released Claude Code 2.1.294 on October 8, 2026. It fixes prompt and agent hooks phrased as instructions, which could allow exactly the actions they were meant to stop. Version 2.1.295 followed the same day with an onFailure setting that makes a crashed or timed-out command or HTTP hook block the action. Releases 2.1.292 and 2.1.293 from earlier in the week closed several permission and sandbox gaps.
Claude Code Releases (GitHub)verified