memorize — Claude Code skill: what it does, how to install
Transfers insights from agent reflections and critique sessions into CLAUDE.md. Keeps your context file sharp by structuring lessons instead of bloating it with raw logs.
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a claude-codeWhat memorize does
The skill covers the curation step within the Agentic Context Engineering workflow. When conversations or reflection tools like /reflexion:critique uncover a pattern, memorize turns that finding into a concise rule. If the target scope is ambiguous, the agent asks whether to harvest the last response, a text selection, or an issue report.
Processing runs through structured stages. It gathers reusable patterns and domain constraints, categorizing each item by impact. Before modifying CLAUDE.md, the skill checks existing entries to avoid duplicates. Unsubstantiated claims without code evidence or documentation must stay out of the context file, otherwise the project manual degrades quickly.
If a fresh finding clashes with existing guidance, the skill favors the more specific, evidence-backed rule. The older entry is flagged for consolidation rather than deleted automatically. It organizes outputs under dedicated sections like architecture decisions and verification rules.
What it helps with
- turn review findings into actionable guidelines in
CLAUDE.md - prevent context file bloating with duplicate instructions
- document codebase anti-patterns discovered during debugging
- maintain clear verification rules for recurring tasks
Example prompts for the agent
- Use memorize to save the lessons from our latest refactoring debate into
CLAUDE.md - Run memorize on the critique output and update our testing rules
- Record this performance anti-pattern in
CLAUDE.mdusing memorize
How to install memorize
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a claude-code2Manually: put the files in place
~/.claude/skills/memorize/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/memorize/
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a codex2Manually: put the files in place
.agents/skills/memorize/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a cursor2Manually: put the files in place
.cursor/skills/memorize/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install NeoLabHQ/context-engineering-kit memorize2Manually: put the files in place
.github/skills/memorize/SKILL.md1Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a gemini-cli2Platform CLI
gemini skills install https://github.com/NeoLabHQ/context-engineering-kit --path skills/memorizeManually: put the files in place
.devin/skills/memorize/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about memorize
How do I install memorize in Claude Code?
Run npx skills add NeoLabHQ/context-engineering-kit --skill memorize -a claude-code.
What does the memorize skill do?
It analyzes recent conversation logs, critique outputs, and code decisions to extract reusable rules. The skill then files these actionable points into appropriate sections of CLAUDE.md.
How does memorize resolve conflicting guidelines?
It prioritizes the more specific rule backed by code evidence. The older contradictory guideline gets marked for review rather than removed automatically.
