Repolode
Submit
BayramAnnakov

BayramAnnakov/claude-reflect

A self-learning add-on for Claude Code that captures your corrections and preferences and syncs them into CLAUDE.md and AGENTS.md as persistent memory.

View on GitHub
Stars
1,827
Forks
165
Open issues
10
Last commit
17 days ago

What is it?

claude-reflect is a Claude Code add-on that turns corrections you give Claude into lasting memory. When you say something like "no, use gpt-5.1 not gpt-5", a hook queues the correction, and the /reflect command lets you review and write approved rules into CLAUDE.md or AGENTS.md so later sessions start with them.

It also mines past session history to find repeated tasks and proposes them as reusable slash commands through /reflect-skills. A separate early-access mod version checks short prompts in-session and shows a band above the prompt where you save each rule with one press.

The Python plugin runs on macOS, Linux and Windows and works with Codex, Cursor and other tools that read AGENTS.md.

Key features

  • 01Captures corrections and positive feedback through Claude Code hooks.
  • 02Syncs approved rules into CLAUDE.md and AGENTS.md, including subdirectory and linked docs.
  • 03Discovers repeating session patterns and generates draft slash commands.
  • 04Detects corrections in any language via a semantic check during review.
  • 05Offers deduplication and a queue view to keep memory files tidy.

Use cases

  • Stop repeating the same correction about model names, test commands or coding conventions across sessions.
  • Bootstrap memory from months of existing Claude Code history on a new machine.
  • Turn a task you ask for often, like a daily review or deploy routine, into a reusable command.

Quick start

  1. 1

    Add the marketplace

    claude plugin marketplace add bayramannakov/claude-reflect
  2. 2

    Install the plugin

    claude plugin install claude-reflect@claude-reflect-marketplace
  3. 3

    Restart Claude Code to activate the plugin

  4. 4

    Run /reflect to process queued learnings

    /reflect

Things to know

  • Requires Claude Code CLI and Python 3.6+.
  • The mod variant needs Claude Code 2.1.286 or newer and makes one model call per checked prompt on your own account.
  • Installing both the mod and the Python plugin captures each correction twice.

Topics

#claude-code#ai-agents#memory#claude-md#prompt-engineering#developer-workflow#claude-skills#productivity#self-learning

FAQ

What is claude-reflect?
A self-learning add-on for Claude Code that captures your corrections and preferences and syncs them into CLAUDE.md and AGENTS.md as persistent memory. claude-reflect is a Claude Code add-on that turns corrections you give Claude into lasting memory. When you say something like "no, use gpt-5.1 not gpt-5", a hook queues the correction, and the /reflect command lets you review and write approved rules into CLAUDE.md or AGENTS.md so later sessions start with them.
Is claude-reflect free to use? What license does it have?
claude-reflect is open source under the MIT license. Check the license terms before using it in commercial products.
Is claude-reflect still maintained?
Actively maintained. The last commit to claude-reflect was on Oct 3, 2026.
What can I use claude-reflect for?
Stop repeating the same correction about model names, test commands or coding conventions across sessions. Bootstrap memory from months of existing Claude Code history on a new machine. Turn a task you ask for often, like a daily review or deploy routine, into a reusable command.
How do I install claude-reflect?
Add the marketplace: claude plugin marketplace add bayramannakov/claude-reflect Install the plugin: claude plugin install claude-reflect@claude-reflect-marketplace Restart Claude Code to activate the plugin Run /reflect to process queued learnings: /reflect
Where can I get claude-reflect?
The source code and installation instructions are on GitHub: https://github.com/BayramAnnakov/claude-reflect

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects