eg_cfg
boazlavon/eg_cfg
EG-CFG is an inference-time algorithm that integrates real-time execution feedback into the language model's decoding loop for robust code generation.
49 10No licenseUpdated 2026-02-08
49
FairOverview
EG-CFG (Execution-Guided Line-by-Line Code Generation) is an advanced framework that steers LLM decoding using dynamic runtime feedback. By incorporating execution traces during token generation, the method ensures that produced code is not only syntactically correct but functionally valid, establishing state-of-the-art results on benchmarks like MBPP, HumanEval, and CodeContests.
Capabilities
- ▸Real-time execution feedback integration during decoding
- ▸Multi-agent parallel trial execution and grid search
- ▸Support for both local models and remote inference endpoints
- ▸Handling complex coding benchmarks (MBPP, HumanEval, DS-1000, CodeContests)
Best for
Running state-of-the-art code generation experiments with real-time execution feedback, Optimizing LLM decoding parameters via grid search or config strings, Executing parallel inference workloads using SLURM clusters
Works with
CursorCodex
Evoa Score breakdown= Σ (score × weight)
49Task usefulness20%77 → +15.4
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%90 → +13.5
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%50 → +7.5
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%24 → +2.9
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%15 → +2.3
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%95 → +9.5
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%46 → +3.7
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%50 → +2.5
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total57.2 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
The framework executes generated code dynamically during the decoding loop, which requires a secure sandbox environment to prevent arbitrary code execution vulnerabilities.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.