Logseq Bridge#

Interact with your Logseq graph by directly reading and writing .md files on disk. No plugins, no HTTP APIs — just filesystem access to your Logseq graph directory.

---

How It Works#

text

Agent → Shell → Read/Write local `.md` files → Logseq reloads on next Cmd/Ctrl+R

Pure file operations. No HTTP APIs, no plugin bridges, no port listeners.

---

Prerequisites#

Your Logseq graph is stored as a local directory structure:

text

{graph_root}/
├── journals/       ← Daily notes *.md
├── pages/          ← Knowledge pages *.md
├── assets/         ← Images, attachments
├── logseq/         ← Config files
│   └── config.edn
└── .logseq/        ← Index database (auto-maintained)
    └── graphs/
        └── logseq_local_*.transit

What you need:

1. Logseq installed and opened at least once (so .logseq/ exists)
2. Read/write access to the graph directory on your filesystem
3. Know your graph path

---

Finding Your Graph Path#

Method 1: From Logseq UI#

Open Logseq → Settings → Advanced → Look for "Current graph directory"

Method 2: From the transit database file#

bash

ls ~/.logseq/graphs/*.transit
# e.g.: logseq_local_E%3A+++Users++MyUser++Logseq.transit

The filename is URL-encoded. Decode it to find your path:

bash

# In WSL/Linux, mount the drive and set the path
export LQ="/mnt/c/Users/MyUser/Logseq"

---

Core Operations#

Set the environment variable first:

bash

export LQ="/path/to/your/logseq/graph"

📝 Write to Today's Journal#

bash

DATE=$(date +%Y_%m_%d)
FILE="$LQ/journals/$DATE.md"

cat >> "$FILE" << 'EOF'
- 🦞 Note title
  - Sub-item 1
  - Sub-item 2
EOF

📖 Read a Journal#

bash

cat "$LQ/journals/$(date +%Y_%m_%d).md"
# Or a specific date
cat "$LQ/journals/2026_04_25.md"

📄 Read a Knowledge Page#

bash

cat "$LQ/pages/Page Name.md"

🔍 Search Notes#

bash

# Search all pages
grep -ri "keyword" "$LQ/pages/"

# Search journals
grep -rn "keyword" "$LQ/journals/"

# Search the database index (faster)
strings ~/.logseq/graphs/*.transit | grep -i "keyword" | head -20

📋 Create a New Page#

bash

cat > "$LQ/pages/New Page.md" << 'EOF'
title:: New Page
tags:: tag1, tag2

- Page content
  - Sub content
EOF

📊 List Statistics#

bash

# Total journals
ls "$LQ/journals/" | wc -l

# Total pages
ls "$LQ/pages/" | wc -l

# Latest 5 journals
ls -t "$LQ/journals/" | head -5

---

Markdown Format Reference#

Write notes in Logseq-compatible Markdown:

SyntaxExampleNoteBullet list- contentEvery line starts with - Indent children - child2-space indentBold**text**Standard MarkdownWiki link[[Page Name]]Auto-links pagesTag#tagnameAuto-indexedPropertykey:: valuePage or block-level propertyTODO markerTODO do somethingRenders as taskDONE markerDONE completedRenders as done

---

Known Issues & Limitations#

IssueDescriptionWorkaroundNo real-time refreshLogseq doesn't watch file changesPress Ctrl+Shift+R (or Cmd+Shift+R) to reloadStatic text readsReads raw .md, not block treeParse Markdown yourself for structured dataCloud sync delayGraph on OneDrive/Dropbox may lagWait a few seconds after writingSpecial characters in pathsChinese/Unicode charsWrap paths in double quotes

---

Quick Start#

bash

# 1. Find your graph path in Logseq Settings → Advanced

# 2. Set the path
export LQ="/path/to/your/logseq/graph"

# 3. Write a test note
DATE=$(date +%Y_%m_%d)
echo "- 🎉 Logseq Bridge test successful!" >> "$LQ/journals/$DATE.md"

# 4. Reload Logseq (Ctrl+Shift+R or Cmd+Shift+R)

---

Best Practices#

1. File operations first — stable, zero dependencies
2. Tell user to reload — Logseq won't auto-detect file changes
3. Date format — Journal filenames are always YYYY_MM_DD.md
4. Page naming — The filename IS the page name; Chinese OK
5. Heredoc — Use << 'EOF' for multi-line content

---

References#

• Logseq: https://logseq.com
• Logseq Markdown Guide: https://docs.logseq.com
• ClawHub: https://clawhub.ai