Developer Criticizes Codex Subagent Implementation: Memory Bloat and Crashes
mertdumenci · x · 2026-07-26
A developer posted a thread criticizing the terrible implementation of subagents in the Codex harness. While praising other features, they noted that subagent behavior is completely opaque, with no clarity on task delegation. Worse, subagents continuously accumulate, driving memory usage up to 45GB and causing crashes, while advanced models can drain 50% of the weekly quota on a single task.
More from coding & agent
- Shopify design lead on AI-era workflow: designers are becoming full-stack editors — KlausCodes · 2026-10-26
- Ultrametric open-sources Open Startup Repo to let AI agents run startup operations — garrytan · 2026-10-26
- MCP fault injection study: agent task scores drop 60% under realistic tool failures, scale doesn't help — dair_ai · 2026-10-26
- General agents burn ~140M tokens/month on LinkedIn outreach; scripts do it 60x cheaper — ChrisHarpon2 · 2026-10-26
- OpenAI upgrades Decisions API: 400 questions per call, native voice and image support — OpenAIDevs · 2026-10-26
- NLI alone rejects 5/6 new agent memories; a 0.45 similarity gate fixes it all — Alternative_Pin9598 · 2026-10-26