AI Engineeringhttps://ai.devops-monk.com/A 23-chapter handbook for Claude Code — permissions, context, skills, hooks, MCP and agents, worked out end to end for engineers who have to ship with it.en-usabhaypratap3537@gmail.com (Abhay)abhaypratap3537@gmail.com (Abhay)Sat, 05 Sep 2026 23:56:44 +0000Permission Modeshttps://ai.devops-monk.com/2026/09/permission-modes/Sat, 05 Sep 2026 12:00:00 +0000https://ai.devops-monk.com/2026/09/permission-modes/abhaypratap3537@gmail.com (Abhay)FundamentalsThe six permission modes, what each one auto-approves, the auto-mode classifier's default block and allow lists, the thresholds at which it stops trusting itself, and the protected and critical paths no mode will auto-approve.Permission Modes

The six permission modes, what each one auto-approves, the auto-mode classifier's default block and allow lists, the thresholds at which it stops trusting itself, and the protected and critical paths no mode will auto-approve.

Read the full article →

]]>Three Ways to Talk to Claude Codehttps://ai.devops-monk.com/2026/09/three-ways-to-talk-to-claude-code/Sat, 05 Sep 2026 11:00:00 +0000https://ai.devops-monk.com/2026/09/three-ways-to-talk-to-claude-code/abhaypratap3537@gmail.com (Abhay)FundamentalsClaude Code accepts input through three channels — CLI flags, slash commands and in-prompt sigils — distinguished by when they are parsed and whether they reach the model. Full flag and command reference, shell-mode semantics, and the input handling layer.Three Ways to Talk to Claude Code

Claude Code accepts input through three channels — CLI flags, slash commands and in-prompt sigils — distinguished by when they are parsed and whether they reach the model. Full flag and command reference, shell-mode semantics, and the input handling layer.

Read the full article →

]]>What Claude Code Actually Ishttps://ai.devops-monk.com/2026/09/what-claude-code-actually-is/Sat, 05 Sep 2026 10:00:00 +0000https://ai.devops-monk.com/2026/09/what-claude-code-actually-is/abhaypratap3537@gmail.com (Abhay)FundamentalsThe execution model behind Claude Code: a language model with no I/O, a harness that supplies tools, and a loop that runs until the task verifies. Plus the tool table and the permission asymmetry it exposes.What Claude Code Actually Is

The execution model behind Claude Code: a language model with no I/O, a harness that supplies tools, and a loop that runs until the task verifies. Plus the tool table and the permission asymmetry it exposes.

Read the full article →

]]>