SAI-baseversion
lutzfinger/SAI-baseversion
SAI is a framework for running personal AI automations with a strong emphasis on evaluation data, cascade execution, and policy gates. It separates workflow design from execution to ensure regression-free automation.
29 14Apache-2.0Updated 2026-07-17
73
GoodOverview
SAI provides a secure, local execution runtime for personal AI workflows by treating evaluation datasets as first-class citizens. The framework features cascade execution (rules, classifier, local LLM, cloud LLM, human), strict policy gates for side effects, append-only audit logging, and hash-verified skill manifests. It requires workflows to declare explicit evaluation data before registration, ensuring reliability across updates.
Capabilities
- ▸Cascade execution across local and cloud LLM tiers with early-stop confidence
- ▸EvalDataset abstraction enforcing canary, edge-case, and workflow verification
- ▸Hash-verified skill loading and manifest validation
- ▸Policy-gated side-effect execution and append-only auditing
- ▸Slack-based feedback and operator-tunable proposal integration
Best for
Automating personal email tagging and triage with a feedback loop, Running custom local AI workflows with strict security and permission scopes, Managing regression-tested personal AI automations with structured evaluation datasets
Works with
Claude CodeCursor
Evoa Score breakdown= Σ (score × weight)
73Task usefulness20%88 → +17.6
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%100 → +15.0
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%39 → +4.7
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%50 → +7.5
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%85 → +8.5
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%44 → +3.5
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 total72.8 / 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
Implements strict policy gates before any side effects, uses fine-grained per-workflow OAuth scopes instead of a superuser token, and maintains an append-only audit log. Reflection mechanisms are strictly bounded and cannot auto-apply code or policy changes.
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.