cd /news/ai-research/openai-s-swe-bench-harness-relies-on… · home › topics › ai-research › article
[ARTICLE · art-130030] src=github.com ↗ pub= topic=ai-research verified=true sentiment=↓ negative

OpenAI's SWE-bench harness relies on unisolated host Docker sockets

OpenAI's SWE-bench evaluation harness relies on unisolated host Docker sockets, according to a bug report filed against the project. The report details errors encountered while running the harness on Docker Desktop with 8 CPUs on an M3 chip and a 15.93 GB memory limit, despite 300 GB of available disk space, and states the evaluation failed to run as expected.

by read1 min views1 publishedSep 15, 2026
OpenAI's SWE-bench harness relies on unisolated host Docker sockets
Image: Michielbdejong (auto-discovered)

Describe the bug

Ran into a few errors

I am especially not sure why I get the second one, i have 15.93 gb allocated memory and 300gb space on my disk.

Steps/Code to Reproduce

Using docker desktop with 8 CPUs available (M3 chip) and 15.93gb memory limit

Expected Results

evaluation is ran

Actual Results

source & further reading
github.com — original article Brig: Run coding agents in a MicroVM sandbox Show HN: Karen – app feedback as easy as complaining A Blender BPY skill that helps smaller LLMs build better 3D models
── more in #ai-research 4 stories · sorted by recency
── more on @openai 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/openai-s-swe-bench-h…] indexed:0 read:1min 2026-09-15 · —