notes-know-you

通过你的印象笔记，让 AI 真正了解你。

This skill syncs your Evernote/Yinxiang notebooks to local Markdown files, analyzes
the content, and updates the agent's USER.md and Memory files with structured insights —
so your AI assistant always has the right context about who you are.

Configuration

Set the following environment variables before using this skill:

Usage

CODEBLOCK0

Step-by-Step Instructions

Step 0 — Check Configuration

1. 1. Verify NOTES_DB_PATH is set and the file exists.
2. If not set, ask the user:

1. 3. If the database doesn't exist yet, guide them through first-time setup (see references/setup.md).
2. Determine:

Step 1 — Sync from Evernote

Run:
CODEBLOCK1

If rate limited (CRITICAL: Rate limit reached): stop and tell the user to wait 25–30 minutes before retrying.

If auth fails: guide the user to get a developer token:

• - Yinxiang: visit https://app.yinxiang.com/api/DeveloperToken.action while logged in
• International: visit INLINECODE27

Then update auth:
CODEBLOCK2

Step 2 — Export to ENEX

CODEBLOCK3

This creates one .enex file per notebook under enex_dir.

Step 3 — Convert ENEX to Markdown

For every .enex file found recursively under enex_dir, run:
CODEBLOCK4

This produces one merged Markdown file per notebook (e.g., Daily.md, Books.md).

Keep a list of all generated Markdown files and their notebook names.

Step 4 — Analyze Notes Content

Read the generated Markdown files. Notes can be large — skim for structure first, then read
recent entries (last 60 days) in full and older entries at a summary level.

Extract insights by notebook type:

Diary / Daily notes (filenames containing: Daily, Diary, Journal, 日记, 每日, 每天):

• - Daily routines and recurring habits
• Emotional patterns and frequent concerns
• Goals and aspirations mentioned
• Recent life events (last 30 days) — record with approximate dates

Work / Project notes (Work, Project, Todo, Tasks, 工作, 项目):

• - Current active projects and their status
• Professional skills and tools used
• Goals and deadlines

Personal / People notes (人, People, Friends, Family, 朋友, 家人):

• - Key relationships and social context
• Personal values and beliefs expressed

Reading / Clippings (Books, 我的剪藏, Clippings, Reading, 读书):

• - Topics and domains of sustained interest
• Preferred content formats (books, articles, videos)

General / Other notebooks:

• - Any recurring themes or topics
• Explicit preferences or opinions stated

Step 5 — Update USER.md

Locate the memory directory:

1. 1. Check INLINECODE34
2. Check .openclaw/ in the current project root
3. Check INLINECODE36
4. Fall back to creating .memory/ in the current directory

Read existing USER.md if it exists. Merge, do not overwrite — preserve facts not
contradicted by the notes.

Write (or update) USER.md with this structure:

CODEBLOCK5

Step 6 — Write Memory Files

For each significant, specific insight, create a memory file in the memory directory.

Use the formats below, saved as {type}_{topic}.md:

User memory (personal attributes, preferences, background facts):
CODEBLOCK6

Project memory (ongoing projects, goals with context):
CODEBLOCK7

Feedback memory (explicit preferences the user has expressed in notes):
CODEBLOCK8

After writing all memory files, update MEMORY.md (the index file) with one line per
new or updated memory:
CODEBLOCK9

Step 7 — Report to User

Summarize what was done:

CODEBLOCK10

Cron Setup

When the user runs /notes-know-you setup-cron [interval]:

1. 1. Default interval: INLINECODE43
2. Parse the interval (e.g., 6h, 12h, 24h, 7d)
3. Use the agent's cron/scheduler to register a recurring job that runs INLINECODE48
4. Confirm to the user: "Scheduled notes-know-you to run every {interval}. Next run: {datetime}."

Privacy Note

All note content stays local. Nothing is sent to external services beyond passing note
text through the AI model's context window for analysis. ENEX files and Markdown files
remain on your machine.