Skip to content

Explore Compare Saved

DevHubBuilt for curious developers. Data from GitHub.

Back to repositories

Public repository / lzjever

air-gapped

Portable Codex CLI + vLLM for air-gapped use

18Stars

2Forks

0Open issues

0Subscribers

Language distribution

Share of source bytes reported by GitHub

Shell61%

Go26%

Python10%

Makefile3%

Recent activity

Latest commits on the default branch

Release v0.5.0: Qwen 3.5 latest (千问 3.5 最新版) as default modellzjever · Mar 2, 2026 Fix vLLM test: remove VLLM_EXTRA_ARGS override, use start-vllm.sh auto-detectionlzjever · Mar 2, 2026 Fix vLLM: use HF safetensors model, auto-detect GPU context, fix OOMlzjever · Mar 2, 2026 Sync Codex config model_context_window with actual backend context sizelzjever · Mar 2, 2026 Unify CPU/GPU defaults, auto-detect resources, enhance tests & docslzjever · Mar 2, 2026 Add codex-proxy (default), vLLM GPU path, streaming fix, testslzjever · Mar 2, 2026 Refactor: llama.cpp binary download, tests, 0.4.0lzjever · Mar 1, 2026 Release 0.2.0: CPU-first defaults, GGUF model, bootstrap GPU opt-inlzjever · Mar 1, 2026 docs: shorten README copy—punchy pain + payoff, one paragraphlzjever · Mar 1, 2026 docs: integrate promotional copy into README (CN/EN)lzjever · Mar 1, 2026

Project details

Created Feb 2026 Updated Mar 2026 MIT License

Contributors

1

lzjever14 commits