{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidzd55y5euxsyxtd5nilspgfcts6ipnds46udnrntc6qiqw6kd2da",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mltlsbbmzj72"
  },
  "path": "/t/larger-context-length-please/1380603#post_3",
  "publishedAt": "2026-05-14T19:21:28.000Z",
  "site": "https://community.openai.com",
  "textContent": "I mainly use Codex with IDA Pro + IDAssitMCP through Visual Studio Code. I’m using markdown for all documentation, although sometimes I feed Codex screenshots if PDF tables aren’t correct in the markdown. Once it reads enough of whatever is loaded in IDA, I will have it create python scripts for data manipulation, working with USB, ethernet and other protocols. And then that feeds back into more questions, more IDA and the cycle repeats. I do try to be context conscious though by doing things like packet captures on specific ports instead of everything for an IP for example.\n\nI understand the model is what it is, I guess I just wanted to throw my vote in the bucket so hopefully if enough people bring it up, OpenAI will release an updated Codex model with larger context. So far I’m loving Codex though and I often use it to cross-check Opus and Sonnet.",
  "title": "Larger context length please"
}