11.0 Enforcing Agent Boundaries | AI-Native Software Engineering — Asha

Turn agent intentions into observable boundaries. Asha explains authority paths, least privilege, credentials, connectors, worktrees versus isolation, a small container experiment, artifact-bound approvals, audit evidence, incident recovery and accountable roles. This standalone overview is chapter 00 of lesson 11 in AI-Native Software Engineering. The following chapters provide deeper worked examples. CHAPTERS 00:00 An instruction is not a locked door 00:20 Connect governance to everyday engineering 00:30 Start with assets and consequences 00:50 Map the actual execution path 01:10 Apply least privilege to the task 01:31 Do not confuse a connector with a control 01:41 Separate working copies from execution isolation 02:11 Build a small observable boundary experiment 02:42 State what the experiment did not prove 03:02 Make network policy concrete 03:22 Require approval for a specific action 03:52 Make the policy executable 04:23 Keep decisions observable 04:43 Design a stop and recovery path 04:53 Prompt for a boundary review 05:24 Package a control verification skill 05:44 Adapt controls to the actual product 05:54 Compare two restrictions 06:14 Keep accountability explicit 06:24 Continue with the practical chapters COURSE AND SOURCES https://storage.googleapis.com/intera... https://docs.docker.com/engine/contai... https://docs.docker.com/engine/networ... https://docs.docker.com/engine/security/ https://docs.docker.com/engine/contai... https://git-scm.com/docs/git-worktree https://modelcontextprotocol.io/speci... https://code.claude.com/docs/en/best-... https://cursor.com/blog/plan-mode https://learn.chatgpt.com/docs/prompt... The container experiment uses generated classroom data only. It verifies selected read/write behavior and loopback-only interfaces, not general escape resistance or confidentiality. Approval matching is a trusted-input teaching fixture, not a production authorization service. Product behavior and supported configuration can change; verify the installed version and official documentation. Created with a fictional AI-generated presenter and illustrative cinematic scenes. Exact diagrams and code are authored educational graphics. Demonstration only — verify with the source. #AINativeEngineering #CodingAgents #SoftwareEngineering