Teaching AI to Reflect

Prologue: 1,754 Perfect Green Lights The team watched the test panel late on the night Aristotle v1.6.0 shipped. Green indicators lit up like dominoes. Python side: 1,166 assertions. TypeScript side: 588 checks. Total: 1,754 automated checks. All green. In code terms, that’s like having cameras and infrared sensors on every wall. A fly couldn’t sneak through without setting off alarms. The team leaned back. The system looked like an iron fortress. ...

TL;DR: Five constraints shaped the Watchdog-Intervention Bridge’s cross-language architecture. Watchdog must intercept LLM tool calls synchronously, so it runs in TypeScript. Intervention must reuse the existing reflection engine and rule system, so it stays in Python. The Bridge adds zero new infrastructure, so it uses subprocess. Communication can’t block every tool call, so batching replaces real-time streaming. MCP’s subprocess model already handles failure, so cross-language risk stays contained. Each decision was the least bad option under the circumstances. ...

TL;DR: Aristotle v1.6.0 introduces the Watchdog-Intervention Bridge, shifting from “reflect after the fact” to “intercept in real time.” A TypeScript watchdog detects 21 signal types before and after tool calls. A Python intervention layer handles 13 violation types, connected via a subprocess bridge. MCP tools expand from 10 stubs to 25 full implementations. Two known bugs remain. Open source on GitHub, MIT license. A Hypothesis Overturned From v1.0 to v1.5, Aristotle answered one question: when AI makes a mistake, how do you make it remember and not repeat it? ...

Five articles in. Time to step back and look at the path itself. Aristotle: Teaching AI to Reflect on Its Mistakes covered the design philosophy and initial implementation. claude-code-reflect: Same Metacognition, Different Soil told the story of porting across platforms. Trust Boundaries: One Idea, Two Systems proposed a trust tiering model. From Scars to Armor: Harness Engineering in Practice validated the theory through refactoring. A Markdown’s Three Lives: From Static Rules to a Git-Backed MCP Server evolved the rule storage from append-only to the GEAR protocol. ...

The previous article, From Scars to Armor: Harness Engineering in Practice, ended with Aristotle having a streamlined router (SKILL.md compressed from 371 lines to 84), an on-demand progressive disclosure architecture, and a working reflect→review→confirm workflow. But one thread never got pulled: Where do confirmed rules actually live? This article follows that thread. It wasn’t planned from the start. Three concrete problems in actual use forced the design out, step by step. ...

Three articles in. Back to code, and a hard look in the mirror. The first post, Aristotle: Teaching AI to Reflect on Its Mistakes, covered the design philosophy and a smooth implementation: three commits in one go. The second, claude-code-reflect: Same Metacognition, Different Soil, described the adaptation cost of moving the same philosophy to Claude Code: continuous iteration from V1 to V3. The third, Trust Boundaries: The Same Idea on Open and Closed Platforms, proposed a tiered trust model and a harness engineering framework. ...

Fundamentum autem est iustitiae fides, id est dictorum conventorumque constantia et veritas. — Cicero, De Officiis The foundation of justice is fides — constancy and truthfulness in words and agreements. The first two posts told the story of two projects. Aristotle: Teaching AI to Reflect on Its Mistakes runs on OpenCode: three commits, done. claude-code-reflect: Same Metacognition, Different Soil runs on Claude Code: V1 through V3, hitting walls the entire way. Both projects solve the same problem: teaching AI agents to learn from mistakes and persist those lessons as durable rules. But the implementation process made me realize there’s a question more important than technology choice. When to trust AI’s judgment, and when to step in? ...

Same metacognitive ability, different soil. The growing patterns look nothing alike. My previous post, Aristotle: Teaching AI to Reflect on Its Mistakes, had three core principles: immediate trigger, session isolation, human in the loop. These sound platform-agnostic. But when I moved the same philosophy to Claude Code, I discovered something: platform differences are much larger than expected. First Hurdle: Plugin System Differences Claude Code’s plugin and OpenCode’s skill are completely different systems. Just getting the plugin installed and recognized took several rounds of struggle. ...

“Knowing yourself is the beginning of all wisdom.” — Aristotle Every time I work with an AI coding assistant, I run into the same problem. Mistakes that were corrected get repeated in the next session. The model isn’t stupid. There’s a structural gap in memory. For example. Last week I corrected a mistake the model made. It apologized, I accepted, and kept working. Today I started a new session, and the same mistake appeared again. ...