### Clone and Install cc-llm4zotero-adapter

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Clone the companion bridge repository, install its dependencies, build the project, and start the bridge server.

```bash
git clone https://github.com/jianghao-zhang/cc-llm4zotero-adapter.git
cd cc-llm4zotero-adapter
npm install
npm run build
npm run serve:bridge
```

--------------------------------

### Install macOS Background Daemon

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Installs the LaunchAgent for running the bridge in the background on macOS.

```bash
./scripts/install-macos-daemon.sh
```

--------------------------------

### Custom Filename Pattern Example 2

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

This example demonstrates another user-customizable filename pattern, incorporating year, first author, and notetitle.

```markdown
## Your customizations

Filename pattern: `{year}-{firstauthor}-{notetitle}.md`
Example: 2020-Buschman-figure-1.md
```

--------------------------------

### Install Codex CLI

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Install the Codex CLI globally using npm. For macOS, Homebrew is also an option. Ensure correct installation path for Windows users.

```bash
npm install -g @openai/codex
```

```bash
brew install --cask codex
```

--------------------------------

### Verify Claude Code CLI Installation

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Run this command to ensure the Claude Code CLI is installed and accessible in your PATH.

```bash
claude
```

--------------------------------

### Custom Filename Pattern Example 1

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

Users can customize the filename pattern by adding a 'Your customizations' section to the skill file. This example shows a pattern using citekey and notetitle.

```markdown
## Your customizations

Filename pattern: `{citekey}-{notetitle}.md`
Example: Buschman2020-figure-1.md
```

--------------------------------

### Embedding Images with Relative Paths

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

This example demonstrates how to correctly format markdown for embedding images using relative paths. It shows the note path, image path, note folder, computed relative path, and the final markdown to be written. Ensure the relative path is correctly calculated to point from the note's directory to the image file.

```markdown
Note path:    {vault}/Logs/paper-notes/Nili2014.md
Image path:   {vault}/Logs/imgs/Nili2014/figure-2.jpg
Note folder:  {vault}/Logs/paper-notes/
Relative:     ../imgs/Nili2014/figure-2.jpg
Write:        ![Figure 2. RSA toolbox schematic](../imgs/Nili2014/figure-2.jpg)
```

--------------------------------

### Manage Bridge Daemon Commands

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Useful commands for managing the bridge daemon's status, start, stop, restart, and uninstall operations.

```bash
npm run daemon:status
npm run daemon:start
npm run daemon:stop
npm run daemon:restart
npm run daemon:uninstall
```

--------------------------------

### Overview Paper Read

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `paper_read` with `mode:'overview'` to get a general understanding of selected papers. This is a preliminary step before deeper reading.

```python
paper_read({
  mode:'overview',
  targets:[...]
})
```

--------------------------------

### Write Note to File using file_io

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

This snippet demonstrates the core logic for writing a note to a file. It constructs the file path and then calls the `file_io` action with the 'write' action type, the resolved file path, and the note content. Error handling for write failures is also included.

```markdown
1. Construct the file path: `{defaultTargetPath}/<filename>.md` unless the user explicitly specifies another folder, using the native path separator from the runtime platform section.
2. Call `file_io({ action:'write', filePath, content:noteContent })`.
3. If writing fails, report the error clearly with the attempted path.
```

--------------------------------

### Template for General Notes

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

Use this template for general notes, topic summaries, or free-form thoughts. It requires a title, creation date, and tags.

```markdown
---
title: "{{noteTitle}}"
created: {{created}}
tags: [zotero]
---

# {{noteTitle}}

{{content}}

---

Written by LLM-for-Zotero.

```

--------------------------------

### Log in to Codex CLI

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Log in to the Codex CLI to authenticate. Your credentials will be saved to a local JSON file.

```bash
codex login
```

--------------------------------

### Read Manifest File

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/analyze-figures.md

Reads the manifest.json file from the MinerU cache directory to access metadata about figures and sections within a paper.

```javascript
file_io({ action:'read', filePath:'{mineruCacheDir}/manifest.json' })
```

--------------------------------

### Import by Identifiers

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Directly imports papers using a list of identifiers (DOI, arXiv ID, ISBN, URL) without prior resolution. This is efficient when identifiers are readily available.

```javascript
library_import({ kind:'identifiers', identifiers:[...] })
```

--------------------------------

### Aggregate Papers from Entire Library

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `zotero_script` to aggregate candidate papers from the entire Zotero library for a literature review. This follows a similar pattern to `library-analysis`.

```python
zotero_script({
  mode:'read',
  description:'Summarize candidate papers for a literature review',
  script:'...'
})
```

--------------------------------

### Search Zotero Collections

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `library_search` to find a specific collection by name, then use `library_retrieve` to search within that collection for review papers.

```python
library_search({
  entity:'collections',
  mode:'search',
  text:'<collection name>'
})
```

--------------------------------

### Analyze Library Statistics with Zotero Script

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/library-analysis.md

This JavaScript code snippet demonstrates how to aggregate statistics for a Zotero library. It iterates through all regular items, counting them by year, type, and tag, and logs the results as JSON. This is useful for generating overviews when the user requests counts or distributions.

```javascript
const items = await Zotero.Items.getAll(env.libraryID, false, false, false);
const byYear = {};
const byType = {};
const byTag = {};
let total = 0;
for (const item of items) {
  if (!item.isRegularItem()) continue;
  total++;
  const year = String(item.getField("date") || "").slice(0, 4) || "unknown";
  byYear[year] = (byYear[year] || 0) + 1;
  byType[item.itemType] = (byType[item.itemType] || 0) + 1;
  for (const tag of item.getTags()) {
    byTag[tag.tag] = (byTag[tag.tag] || 0) + 1;
  }
}
env.log(JSON.stringify({ total, byYear, byType, byTag }, null, 2));
```

--------------------------------

### Read References Section from MinerU Cache

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Reads the references section from a paper's MinerU cache. Requires the manifest.json to find character offsets and then reads the full.md file.

```javascript
file_io({ action:'read', filePath:'{mineruCacheDir}/manifest.json' })
file_io({ action:'read', filePath:'{mineruCacheDir}/full.md', offset:<charStart>, length:<charEnd - charStart> })
```

--------------------------------

### Import Multiple Papers by DOIs

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Imports multiple papers into the Zotero library by providing a list of their DOIs in a single call. This enables batch importing.

```javascript
library_import({ kind:'identifiers', identifiers:['<DOI1>', '<DOI2>', ...] })
```

--------------------------------

### Including Figures in Zotero Notes

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

This snippet shows how to embed figures within Zotero notes using a file path. The `note_write` tool automatically handles the import of these file-based images.

```markdown
![Caption](file:///{mineruCacheDir}/images/filename.png)
```

--------------------------------

### Import Single Paper by DOI

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Imports a single paper into the Zotero library using its DOI. This is used when only one paper needs to be added.

```javascript
library_import({ kind:'identifiers', identifiers:['<DOI>'] })
```

--------------------------------

### Read Figure Image File

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/analyze-figures.md

Reads a figure image file directly from the MinerU cache using its relative path. This allows visual models to analyze the image content.

```javascript
file_io({ action:'read', filePath:'{mineruCacheDir}/<figure_path>' })
```

--------------------------------

### Import with Target Collection

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Imports papers into a specific Zotero collection by including the targetCollectionId. This allows for organized imports.

```javascript
library_import({ kind:'identifiers', identifiers:['<DOI1>', '<DOI2>', ...], targetCollectionId:'<collection ID>' })
```

--------------------------------

### Search Zotero Library for Literature Review

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `library_retrieve` to search Zotero metadata or indexed text for papers relevant to a given topic. This is useful for broad searches when translation or terminology variants might improve recall. Use `intent:'summarize'` for collection-grounded taxonomies.

```python
library_retrieve({
  query:'<topic>',
  queryVariants:[...],
  intent:'enumerate',
  depth:'metadata'|'evidence'
})
```

```python
library_retrieve({
  scope:{ collectionIds:[<collectionId>] },
  query:'<review topic>',
  queryVariants:[...],
  intent:'enumerate',
  depth:'metadata'|'evidence'
})
```

```python
library_retrieve({
  scope:{ collectionIds:[<collectionId>] },
  query:'<review topic>',
  queryVariants:[...],
  intent:'summarize'
})
```

--------------------------------

### Template for Paper Notes

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/write-note.md

Use this template for notes related to specific academic papers. It includes fields for citation, metadata, and references.

```markdown
---
title: "{{paperTitle}}"
created: {{created}}
citekey: {{citekey}}
doi: {{doi}}
journal: {{journal}}
year: {{year}}
tags: [zotero]
---

# {{paperTitle}}

## Summary
Brief overview of the paper's main contribution and what problem it addresses.

## Key Findings
- The most important results, conclusions, or contributions of the paper.

## Methodology
Summary of the research methodology, experimental setup, or analytical approach.

## My Notes
Personal thoughts, critiques, open questions, and connections to other work.

## References
{{fullCitation}}

---

Written by LLM-for-Zotero.

```

--------------------------------

### Visual Paper Read

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/analyze-figures.md

Uses the paper_read tool in visual mode to find pages containing a figure or table and retrieve its image for analysis when MinerU cache is unavailable.

```javascript
paper_read({ mode:'visual', query:'<figure/table label>' })
```

--------------------------------

### Visual Paper Read

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `paper_read` with `mode:'visual'` or MinerU `file_io` to read image data from papers when figures are directly relevant to the review's findings.

```python
paper_read({
  mode:'visual',
  query:'...'
})
```

--------------------------------

### Read References Section Without MinerU Cache

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

If MinerU cache is not available, this method reads the references section using a targeted paper read query.

```javascript
paper_read({ mode:'targeted', query:'reference number or References section' })
```

--------------------------------

### Check Bridge Health

Source: https://github.com/yilewang/llm-for-zotero/blob/main/README.md

Verify that the Zotero Claude bridge is running and accessible by checking its health endpoint.

```bash
curl -fsS http://127.0.0.1:19787/healthz
```

--------------------------------

### Resolve DOI using Title and Author

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

If resolving by title alone fails, this method attempts to resolve the DOI by including the first author's name along with the title.

```javascript
literature_search({ workflow:'answer', mode:'metadata', title:'<title>', author:'<first author>' })
```

--------------------------------

### Targeted Paper Read

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/analyze-figures.md

Uses the paper_read tool in targeted mode to retrieve the text of a figure/table label and its surrounding discussion when MinerU cache is unavailable.

```javascript
paper_read({ mode:'targeted', query:'<figure/table label and surrounding discussion>' })
```

--------------------------------

### Search for Vague Description

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Uses a literature search to find a paper based on a vague description, including title and author information. This is useful when direct identifiers are not provided.

```javascript
literature_search({ workflow:'answer', mode:'search', query:'...', author:'...' })
```

--------------------------------

### Targeted Paper Read

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/literature-review.md

Use `paper_read` with `mode:'targeted'` to find specific information or answer focused questions within selected papers. This is useful for extracting details about methods or key findings.

```python
paper_read({
  mode:'targeted',
  query:'...', 
  targets:[...]
})
```

--------------------------------

### Read Section Text from MinerU Cache

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/analyze-figures.md

Reads a specific section's text from the MinerU cache using character offsets. This is useful for retrieving figure captions and surrounding discussion.

```javascript
file_io({ action:'read', filePath:'{mineruCacheDir}/full.md', offset:<charStart>, length:<charEnd - charStart> })
```

--------------------------------

### Resolve DOI using Exact Title

Source: https://github.com/yilewang/llm-for-zotero/blob/main/src/agent/skills/import-cited-reference.md

Resolves a paper's DOI from CrossRef/Semantic Scholar using its exact title. This is the primary method for DOI resolution.

```javascript
literature_search({ workflow:'answer', mode:'metadata', title:'<exact title>' })
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.