Capture session learnings into the right Claude Code memory tier. /notation:notate grounds itself in the session transcript and project profile, turns each learning into a routed proposal (global CLAUDE.md, ~/.claude/notes/, per-project MEMORY.md, or project docs), and applies them through an interactive picker - with a lean bias that keeps CLAUDE.md small. /notation:notate-all applies every proposal without the picker. Ships a notation-audit skill that flags bloat, stale notes, and orphaned index entries.
No model invocation
Executes directly as bash, bypassing the AI model
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Capture this session's learnings, route each to the right memory tier, and apply every proposal automatically - no interactive picker. Same routing and lean-CLAUDE.md bias as /notation:notate, just without the strategy question. Still prints proposals and backs up CLAUDE.md.
Capture this session's learnings and route each one to the right memory tier - global CLAUDE.md, ~/.claude/notes/, per-project MEMORY.md, or project docs. Grounds proposals in the actual project + session, then applies them through an interactive picker. Keeps CLAUDE.md lean.
A Claude Code plugin for keeping your AI memory tidy and well-placed.
Most "update CLAUDE.md" commands dump everything into one file. notation does the opposite: it routes each learning to the tier where it belongs, with a deliberate bias toward keeping CLAUDE.md lean so it does not bloat the every-session prompt.
| Tier | Location | For |
|---|---|---|
| Global rules | ~/.claude/CLAUDE.md | High-frequency, cross-project rules that fire almost every session |
| Topical notes | ~/.claude/notes/*.md | Tool / platform / API-specific gotchas, loaded on demand, indexed in CLAUDE.md |
| Project memory | ~/.claude/projects/<project>/memory/ | Per-project facts, one per frontmatter file, indexed by MEMORY.md |
| Project docs | ./.claude/docs/, ./CLAUDE.md | Team-shared guides and conventions tracked in the repo |
/notation:notateCapture this session's learnings and route each one to the correct tier. It first grounds itself in the real context - scanning the session transcript for learnings that scrolled out of view and reading the project's own CLAUDE.md, MEMORY.md, and recent git history - so each proposal is deduped and correctly aimed at global vs this project. Every learning becomes a structured proposal (tier, rationale, new-or-update, confidence, concrete diff). It then applies them through a plan-mode-style interactive picker: Apply all, Review by tier (multi-select per tier), or Skip all. It backs up CLAUDE.md before editing it and keeps indexes in sync (Topical Notes Index, MEMORY.md pointers).
/notation:notate-all (or /notation:notate all)Same capture and routing as /notation:notate, but it skips the strategy picker and applies every proposal automatically. It still prints the full proposal list with diffs first and still backs up CLAUDE.md before any inline edit - you just are not asked which to apply. Use it when you trust the routing and want a one-shot save at the end of a session.
notation-audit skillAuto-invocable health check. Ask it to "audit my notes" or "check CLAUDE.md for bloat" and it reports inline entries that should move to notes, orphaned index links, oversized notes, memory files missing pointers, cross-tier duplication, and stale backups - then offers targeted fixes.
claude plugin marketplace add ringo380/robworks-claude-code-plugins
claude plugin install notation@robworks-claude-code-plugins
Then run /notation:notate at the end of a work session.
MIT - see LICENSE.
npx claudepluginhub ringo380/robworks-claude-code-plugins --plugin notationDrive Google Colab from Claude Code over a bundled MCP server — open Drive notebooks, run code in the kernel, manage multiple sessions, upload/download VM files, mount Drive, read Colab secrets, inspect variables, edit notebook cells, and run fine-tune round-trip + local MLX conversion helpers. Optional headless/Playwright backend. Includes guided /colab:setup and /colab:help commands.
Multi-agent pull request review that surfaces every finding instead of hiding low-confidence ones. /deep-review:review scores each issue, then presents all of them in-session triaged into now/soon/optional tiers and posts to the PR on confirm. /deep-review:auto skips scoring, auto-applies fixes for every issue to the working tree, and posts a summary comment - no confirmation.
Comprehensive C4 architecture documentation workflow with bottom-up code analysis, component synthesis, container mapping, and context diagram generation
Build and maintain an LLM-curated personal knowledge base in your project — Andrej Karpathy's LLM Wiki pattern, designed to scale to thousands of pages without becoming a context bottleneck. Now with an optional compiled graph layer for typed, provenance-backed relationships.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).