Run any Skill in Manus with one click

Get Started

source-command-reflect

Stars21

Forks0

UpdatedJuly 21, 2026 at 23:52

Convert repeated human corrections into durable repository agent rules.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

thecompez

thecompez/ai-for-modern-cpp

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

SKILL.md

readonly

name	source-command-reflect
description	Convert repeated human corrections into durable repository agent rules.

source-command-reflect

Use this skill when the user asks to run the migrated source command reflect.

Command Template

Reflect

Use this after the human corrected AI-generated work.

Process

Read the current diff.
Identify what the human changed.
Decide whether the correction is local or a general repository rule.
If it is general, update AGENTS.md with a short stable rule.
Route the lesson to any affected task guide under docs/agent/.
If it is reviewable, update docs/REVIEW.md.
Add or update a pattern and eval scenario when they make the behavior clearer.
Extend tests/knowledge_contract.cmake when the invariant is machine-checkable.
Keep executable examples synchronized with the new rule.
Separate first causal failures from cascading IDE or generated-file diagnostics when generalizing the correction.
Do not add noisy or overly specific rules.

Output

Summarize:

What was learned.
Which file was updated.
Why the rule is general enough to keep.
Which guide, review check, pattern, eval, or contract test was synchronized.

More from this repository

same repository

source-command-design-qt-quick-ui

thecompez/ai-for-modern-cpp

Design and implement a Qt Quick interface with a deliberate visual system and a clean C++ module boundary.

2026-07-2221

source-command-implement

thecompez/ai-for-modern-cpp

Implement a requested change using the repository's modern C++ agent loop.

2026-07-2221

source-command-release

thecompez/ai-for-modern-cpp

Prepare a release using a strict verification-first workflow.

2026-07-2221

source-command-test

thecompez/ai-for-modern-cpp

Build and test the repository without making unnecessary code changes.

2026-07-2221

source-command-start-project

thecompez/ai-for-modern-cpp

Enforce the human-approved project-name gate before creating a new product, application, service, library, tool, repository, or implementing an idea. Use when starting new project work or when another workflow is about to generate a project.

2026-07-2221

source-command-add-module

thecompez/ai-for-modern-cpp

Add a modern C++ module with a documented boundary, CMake registration, and tests.

2026-07-2221

name	source-command-reflect
description	Convert repeated human corrections into durable repository agent rules.

source-command-reflect

Use this skill when the user asks to run the migrated source command reflect.

Command Template

Reflect

Use this after the human corrected AI-generated work.

Process

Read the current diff.
Identify what the human changed.
Decide whether the correction is local or a general repository rule.
If it is general, update AGENTS.md with a short stable rule.
Route the lesson to any affected task guide under docs/agent/.
If it is reviewable, update docs/REVIEW.md.
Add or update a pattern and eval scenario when they make the behavior clearer.
Extend tests/knowledge_contract.cmake when the invariant is machine-checkable.
Keep executable examples synchronized with the new rule.
Separate first causal failures from cascading IDE or generated-file diagnostics when generalizing the correction.
Do not add noisy or overly specific rules.

Output

Summarize:

What was learned.
Which file was updated.
Why the rule is general enough to keep.
Which guide, review check, pattern, eval, or contract test was synchronized.