Bayramannakov Claude Reflect
bayramannakov-claude-reflect is a Claude Code plugin for the Operate phase that captures session feedback and syncs it into CLAUDE.md and AGENTS.md for self-learning agent behavior.
by BayramAnnakov · github.com/BayramAnnakov/claude-reflect
Install a self-learning layer that records your corrections and preferences into CLAUDE.md and AGENTS.md across projects.
Add it to Claude Code
Install the plugin in Claude Code. One command, paste-ready.
/plugin install bayramannakov-claude-reflect@BayramAnnakov/claude-reflectBuilt to be called by your agent
Skillselion is itself an MCP server. Your agent can pull this entry and a paste-ready install config straight from the API - no copy-paste.
Retrieve this entry with skillselion.get_details("plugin:BayramAnnakov/claude-reflect") and the paste-ready config with skillselion.get_install_config("plugin:BayramAnnakov/claude-reflect").
What it does
bayramannakov-claude-reflect is a Claude Code plugin that implements a self-learning loop: when you correct the agent or affirm good behavior, reflect captures those signals and writes them into CLAUDE.md and AGENTS.md so later sessions inherit your standards. Solo builders wear every hat, so re-explaining stack choices, tone, or review rules every day is expensive; this plugin targets that friction with a journey-wide process skill posture. The canonical shelf is Operate → iterate because the artifact is living project memory, but invoke it during build, ship review, and even early scope conversations whenever decisions should stick. It is one plugin from BayramAnnakov/claude-reflect in the community category, not a browser MCP or a 45-skill marketplace. Complexity is intermediate: you need Claude Code, readable project instruction files, and discipline to let the sync run after meaningful feedback. Outcomes are fewer regressions in agent behavior and a auditable preference trail in repo-local docs.
Highlights
- Captures corrections, positive feedback, and preferences from Claude Code sessions
- Syncs learned rules into CLAUDE.md and AGENTS.md for continuity
- Self-learning system aimed at reducing repeated instruction churn
- Single-plugin community bundle (~880 GitHub stars in catalog metadata)
- Works as ongoing agent tooling rather than a one-shot code generator
Why builders use it
Every new Claude Code session forgets your corrections until you manually rewrite project instructions again.
Approved preferences and fixes accumulate in CLAUDE.md and AGENTS.md so the agent improves continuity without you repeating the same rules.
At a glance
- Type - Plugin in Productivity.
- Adoption - 0 installs, 880 stars, 0 votes.
FAQ
Who is bayramannakov-claude-reflect for?
Claude Code users who want corrections and preferences automatically reflected in project agent docs.
When should I use bayramannakov-claude-reflect?
Whenever you give repeated feedback, approvals, or style corrections and want them persisted for future agent runs.
How do I add bayramannakov-claude-reflect to my agent?
Register the reflect plugin from BayramAnnakov/claude-reflect in Claude Code and keep CLAUDE.md and AGENTS.md in your project so sync targets exist.
Comments
Share how you use bayramannakov-claude-reflect, gotchas, or tips for other indie builders.
No comments yet - be the first to share how you use it.