reflect

Converts agent errors and user corrections into permanent enforceable rules.

Updated Oct 2, 2025
One-click install
npx skills add https://github.com/YuriiYInno/Innogram --skill reflect-yuriiyinno
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reflect
Source: https://github.com/YuriiYInno/Innogram/tree/main/.codex/skills/reflect
Command: npx skills add https://github.com/YuriiYInno/Innogram --skill reflect-yuriiyinno

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agent mistakes, test failure patterns, and repeated user corrections often get fixed once and then forgotten, causing the same errors to recur. This Skill automates the error-to-rule cycle so every incident becomes a permanent, enforceable rule in your project documentation. ## Core Features & Use Cases - Root Cause Classification: Categorizes incidents into six root causes such as ambiguous instructions, missing context, or missing enforcement. - Structured Rule Drafting: Generates rules with trigger, root cause, MUST/MUST NOT statements, correct and incorrect examples, and enforcement method. - Placement Guidance: Routes each rule to the right file, such as AGENTS.md for global policy, GOLDEN-PRINCIPLES.md for code style, or a specific SKILL.md. - Use Case: After an agent repeatedly creates files without the required naming prefix, run this Skill to draft a mandatory naming rule, get approval, and write it into the governing documentation with a verification pass. ## Quick Start Ask the agent to reflect on the recent mistake where it created files without the required prefix and turn it into a permanent rule.

Frequently Asked Questions about reflect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn an agent mistake into a permanent rule?▼

Describe the incident to the reflect skill, which gathers what happened, classifies the root cause, drafts a rule with correct and incorrect examples, and writes it to the appropriate governance file after your approval.

How to stop an AI agent from repeating the same error?▼

Use the reflect workflow to convert the error into a MUST or MUST NOT rule with an enforcement method such as a hook, skill instruction, or review checklist. The rule is placed where future agent runs will load it.

Where does the reflect skill write new rules?▼

Rules go to AGENTS.md for global policy, .codex/GOLDEN-PRINCIPLES.md for code style, a specific SKILL.md for single-skill issues, or STABILIZATION.md as example cycles. Placement depends on the rule's scope.

Does the reflect skill modify files without approval?▼

No. The skill presents the drafted rule, target file, and exact section, then waits for explicit user approval before writing. Hook modifications also require separate explicit approval.

What root causes does the reflect skill detect?▼

It classifies incidents into six categories: ambiguous instruction, missing context, wrong model level, missing enforcement, missing workflow step, and stale or conflicting documentation.