From 60d558e9a7eb1944fc58ee88e121a5db5248a332 Mon Sep 17 00:00:00 2001 From: tanaka takashi Date: Mon, 10 Aug 2026 01:39:40 +0900 Subject: fix(agent): fail fast on sandbox policy denial Treat Policy Engine denials and UnauthorizedAccessException results as terminal tool failures by adding RetryAllowed metadata to ToolExecutionResult. Stop the Agent loop immediately after recording the denied tool result, transition the Task to failed, and run normal Task resource cleanup instead of returning the same failure to the LLM for another iteration. Correctable validation and process failures retain the existing bounded retry behavior. Expose RetryAllowed in llm.jsonl and document the fail-fast security invariant in Specifications, Design, and ImplementationPlan. Tests: Agent.Core.Tests 50/50 passed with skip 0; Release build passed with 0 warnings and 0 errors. The full Windows suite remained environment-blocked outside an elevated token: 14 tests failed while creating/deriving the fixed AppContainer or creating symlinks; 11 passed and none were skipped. Release: refreshed the win-x64 self-contained folder binaries, publish manifest, ZIP, and checksum while preserving the local agent.yaml. The clean ZIP contains no runtime logs or Task state. --- doc/Specifications.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'doc/Specifications.md') diff --git a/doc/Specifications.md b/doc/Specifications.md index e723d03..cc04762 100644 --- a/doc/Specifications.md +++ b/doc/Specifications.md @@ -231,7 +231,7 @@ Vision入力は設定可能な画像数、1画像あたりのsize、request合 ### 7.2 Tool未使用検出 -AgentはTaskの期待副作用と観測結果を比較し、実行が必要なのにtool callが0件の場合はTaskをcompletedにしない。再試行はAgent Loopの失敗回数へ算入する。純粋な説明・分析Taskにはtool callを強制しない。 +AgentはTaskの期待副作用と観測結果を比較し、実行が必要なのにtool callが0件の場合はTaskをcompletedにしない。再試行はAgent Loopの失敗回数へ算入する。純粋な説明・分析Taskにはtool callを強制しない。Sandbox Policyによる未許可操作は修正可能なTool失敗としてLLMへ再試行させず、初回のdenyでTaskをfailedにしてcleanupを実行する。 Internal Toolが返す読取・検索・tail結果は設定可能な上限を持ち、contextを無制限に消費してはならない。上限値は設計および設定schemaで定義する。 -- cgit v1.2.3