### Install Graphlit MCP Server via Smithery for Claude

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

This command installs the graphlit-mcp-server for Claude Desktop using the Smithery CLI. It automates the setup process for the specified client.

```bash
npx -y @smithery/cli install @graphlit/graphlit-mcp-server --client claude
```

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

### Install Graphlit MCP Server via NPX

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

Use this command to install the graphlit-mcp-server using NPX. This command is applicable for both Windsurf and Cline IDE installations.

```bash
npx -y graphlit-mcp-server
```

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

### askGraphlit

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Queries Graphlit's AI assistant to get code samples and guidance about using the Graphlit API or SDKs.

```APIDOC
## askGraphlit — Ask questions about the Graphlit platform / API

### Description
Queries Graphlit's own AI assistant to get code samples and guidance about using the Graphlit API or SDKs (Python, TypeScript, C#).

### Arguments
- **prompt** (string) - Required - The question to ask the AI assistant.

### Request Example
```typescript
{
  "tool": "askGraphlit",
  "arguments": {
    "prompt": "How do I ingest a PDF from a URL and then search it with a vector query using the Python SDK?"
  }
}
```

### Response Example
Markdown-formatted explanation with code sample.
```

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

### Retrieve Current Project Metadata

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Call `read_resource("projects://")` to get metadata about the current project, including its name, workflow, quota, credits, and token usage.

```typescript
read_resource("projects://")
// {
//   "name": "My Project",
//   "workflow": { "id": "..." },
//   "quota": { "storage": 10737418240, "fileCount": 10000 },
//   "credits": [{ "name": "Prompt completion", "credits": 2.35 }],
//   "tokens":  [{ "name": "Text embedding", "tokens": 48200 }]
// }
```

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

### Manage Feed Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Utilize `read_resource` with `feeds://` to list available feeds or retrieve detailed metadata for a specific feed using its GUID.

```typescript
read_resource("feeds://")
// [{ name: "Slack [engineering]", uri: "feeds://feed-guid" }]
```

```typescript
read_resource("feeds://feed-guid")
// { "id": "...", "name": "Slack [engineering]", "type": "Slack",
//   "readCount": 342, "state": "Enabled", "lastReadDate": "..." }
```

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

### VS Code User Settings Configuration for Graphlit MCP Server

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

Add this JSON block to your VS Code User Settings (JSON) or a workspace-specific .vscode/mcp.json file to configure the Graphlit MCP Server. This setup prompts for necessary credentials.

```json
{
  "mcp": {
    "inputs": [
      {
        "type": "promptString",
        "id": "organization_id",
        "description": "Graphlit Organization ID",
        "password": true
      },
      {
        "type": "promptString",
        "id": "environment_id",
        "description": "Graphlit Environment ID",
        "password": true
      },
      {
        "type": "promptString",
        "id": "jwt_secret",
        "description": "Graphlit JWT Secret",
        "password": true
      }
    ],
    "servers": {
      "graphlit": {
        "command": "npx",
        "args": ["-y", "graphlit-mcp-server"],
        "env": {
          "GRAPHLIT_ORGANIZATION_ID": "${input:organization_id}",
          "GRAPHLIT_ENVIRONMENT_ID": "${input:environment_id}",
          "GRAPHLIT_JWT_SECRET": "${input:jwt_secret}"
        }
      }
    }
  }
}
```

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

### Delete individual resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Delete individual resources by GUID.

```APIDOC
## deleteContent / deleteConversation / deleteCollection / deleteFeed

### Description
Delete individual resources by GUID.

### Method
Not applicable (Tool calls)

### Endpoint
Not applicable (Tool calls)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
{ "tool": "deleteContent",      "arguments": { "id": "content-guid" } }
{ "tool": "deleteConversation", "arguments": { "id": "conv-guid" } }
{ "tool": "deleteCollection",   "arguments": { "id": "collection-guid" } }  // does NOT delete contents
{ "tool": "deleteFeed",         "arguments": { "id": "feed-guid" } }         // DOES delete feed contents
```

### Response
#### Success Response (200)
No specific response body documented for these delete operations.
```

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

### Full mcp.json Configuration with Data Connector Credentials

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

This configuration includes all essential Graphlit environment variables and optional credentials for various data connectors like Slack, Google Email, Notion, and more. Ensure all sensitive information is kept secure.

```json
{
    "mcpServers": {
        "graphlit-mcp-server": {
            "command": "npx",
            "args": [
                "-y",
                "graphlit-mcp-server"
            ],
            "env": {
                "GRAPHLIT_ORGANIZATION_ID": "your-organization-id",
                "GRAPHLIT_ENVIRONMENT_ID": "your-environment-id",
                "GRAPHLIT_JWT_SECRET": "your-jwt-secret",
                "SLACK_BOT_TOKEN": "your-slack-bot-token",
                "DISCORD_BOT_TOKEN": "your-discord-bot-token",
                "TWITTER_TOKEN": "your-twitter-token",
                "GOOGLE_EMAIL_REFRESH_TOKEN": "your-google-refresh-token",
                "GOOGLE_EMAIL_CLIENT_ID": "your-google-client-id",
                "GOOGLE_EMAIL_CLIENT_SECRET": "your-google-client-secret",
                "LINEAR_API_KEY": "your-linear-api-key",
                "GITHUB_PERSONAL_ACCESS_TOKEN": "your-github-pat",
                "JIRA_EMAIL": "your-jira-email",
                "JIRA_TOKEN": "your-jira-token",
                "NOTION_API_KEY": "your-notion-api-key"
            }
        }
    }
}
```

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

### ingestFile

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingest a local file. Reads a file from the local filesystem, base64-encodes it, and ingests it asynchronously. MIME type is detected automatically.

```APIDOC
## ingestFile

### Description
Ingest a local file. Reads a file from the local filesystem, base64-encodes it, and ingests it asynchronously. MIME type is detected automatically.

### Method
Not applicable (Tool call)

### Endpoint
Not applicable (Tool call)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
{ "tool": "ingestFile", "arguments": { "filePath": "/home/user/reports/q4-2024.pdf" } }
```

### Response
#### Success Response (200)
- **id** (string) - The unique identifier for the ingested content.

#### Response Example
```json
{ "id": "content-guid" }
```
```

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

### List and Read Content Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use `read_resource` with `contents://` to list all finished content items or retrieve the full Markdown text and metadata for a specific content item by its ID.

```typescript
// List all finished contents
read_resource("contents://")
// Returns: [{ name, description, uri: "contents://{id}", mimeType }]
```

```typescript
// Retrieve full Markdown text + metadata for one content
read_resource("contents://3f2a1b4c-0000-0000-0000-000000000002")
// Returns Markdown with fields like:
// **Content ID:** ...
// **File Name:** report.pdf
// **Downloadable Original:** https://cdn.graphlit.dev/...
// **Page #1:** ...extracted text...
```

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

### Delete Individual Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Delete specific content, conversations, collections, or feeds by their GUID. Deleting a collection does not delete its contents, but deleting a feed does delete its contents.

```typescript
{ "tool": "deleteContent",      "arguments": { "id": "content-guid" } }
```

```typescript
{ "tool": "deleteConversation", "arguments": { "id": "conv-guid" } }
```

```typescript
{ "tool": "deleteCollection",   "arguments": { "id": "collection-guid" } }  // does NOT delete contents
```

```typescript
{ "tool": "deleteFeed",         "arguments": { "id": "feed-guid" } }         // DOES delete feed contents
```

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

### Enumerate Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Discover valid resource IDs for various services like Slack channels, Notion databases/pages, Discord guilds/channels, SharePoint libraries/folders, Dropbox/Box folders, Linear projects, and Google/Microsoft calendars. Authentication via environment variables is often required.

```typescript
// Slack channels
{ "tool": "listSlackChannels", "arguments": {} }
```

```typescript
// Notion databases
{ "tool": "listNotionDatabases", "arguments": {} }
```

```typescript
// Notion pages inside a database
{ "tool": "listNotionPages", "arguments": { "databaseId": "..." } }
```

```typescript
// Discord guilds then channels
{ "tool": "listDiscordGuilds",   "arguments": {} }
{ "tool": "listDiscordChannels", "arguments": { "guildId": "123456789" } }
```

```typescript
// SharePoint libraries then folders
{ "tool": "listSharePointLibraries", "arguments": {} }
{ "tool": "listSharePointFolders",   "arguments": { "libraryId": "library-guid" } }
```

```typescript
// Dropbox / Box folders
{ "tool": "listDropboxFolders", "arguments": { "folderPath": "/Photos" } }
{ "tool": "listBoxFolders",     "arguments": { "folderId": "0" } }
```

```typescript
// Linear projects
{ "tool": "listLinearProjects", "arguments": {} }
```

```typescript
// Google / Microsoft calendars
{ "tool": "listGoogleCalendars",    "arguments": {} }
{ "tool": "listMicrosoftCalendars", "arguments": {} }
```

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

### Ingest Cloud Storage Files (Dropbox/Box)

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from cloud storage services like Dropbox or Box. Dropbox requires DROPBOX_APP_KEY, DROPBOX_APP_SECRET, DROPBOX_REFRESH_TOKEN. Box requires BOX_CLIENT_ID, BOX_CLIENT_SECRET, BOX_REDIRECT_URI, BOX_REFRESH_TOKEN.

```typescript
// Dropbox
{
  "tool": "ingestDropboxFiles",
  "arguments": { "path": "/Documents/Reports", "readLimit": 100 }
}
// env: DROPBOX_APP_KEY, DROPBOX_APP_SECRET, DROPBOX_REFRESH_TOKEN

// Box
{
  "tool": "ingestBoxFiles",
  "arguments": { "folderId": "123456", "readLimit": 100 }
}
// env: BOX_CLIENT_ID, BOX_CLIENT_SECRET, BOX_REDIRECT_URI, BOX_REFRESH_TOKEN
```

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

### Retrieve Conversation History

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Access conversation history using `read_resource` with `conversations://`. Provide a conversation GUID to retrieve the full message history formatted as Markdown.

```typescript
read_resource("conversations://")
```

```typescript
read_resource("conversations://conv-guid")
// Returns formatted Markdown:
// **Conversation ID:** conv-guid
// user:
// What were the main themes in last week's standups?
// ---
// assistant:
// The main themes were deployment delays and...
// **Cited Source [1]**: contents://...
```

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

### ingestBoxFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from Box. Requires Box credentials.

```APIDOC
## ingestBoxFiles

### Description
Ingests files from Box.

### Arguments
- **folderId** (string) - Required - The ID of the Box folder.
- **readLimit** (number) - Optional - The maximum number of files to read.

### Environment Variables Required
- BOX_CLIENT_ID, BOX_CLIENT_SECRET, BOX_REDIRECT_URI, BOX_REFRESH_TOKEN
```

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

### askGraphlit Tool

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Queries Graphlit's AI assistant for guidance on using the Graphlit API or SDKs. This tool is useful for obtaining code samples and explanations related to Graphlit platform functionalities.

```json
{
  "tool": "askGraphlit",
  "arguments": {
    "prompt": "How do I ingest a PDF from a URL and then search it with a vector query using the Python SDK?"
  }
}
```

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

### configureProject

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Configures the default workflow and LLM specification for the Graphlit project. This tool should only be called when explicitly requested by the user.

```APIDOC
## configureProject — Configure default workflow and LLM specification

### Description
Sets up the Graphlit project's default ingestion workflow (document preparation, entity extraction) and default conversation specification (LLM model, hybrid search, reranking). Should only be called when explicitly requested by the user.

### Arguments
- **modelServiceType** (string) - Required - Specifies the model service to use (e.g., ANTHROPIC, OPEN_AI, GOOGLE).
- **configureConversationSpecification** (boolean) - Optional - Enables conversation specification (e.g., Claude 3.7 Sonnet + hybrid search + Cohere reranking).
- **configurePreparationSpecification** (boolean) - Optional - Enables document preparation specification (e.g., vision-LLM document prep).
- **configureExtractionSpecification** (boolean) - Optional - Enables entity extraction into knowledge graph.

### Request Example
```json
{
  "tool": "configureProject",
  "arguments": {
    "modelServiceType": "ANTHROPIC",
    "configureConversationSpecification": true,
    "configurePreparationSpecification": true,
    "configureExtractionSpecification": true
  }
}
```

### Response Example
```json
{
  "id": "3f2a1b4c-0000-0000-0000-000000000001"
}
```
```

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

### ingestNotionPages

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests pages from a Notion database. Requires NOTION_API_KEY environment variable.

```APIDOC
## ingestNotionPages

### Description
Ingests pages from a Notion database.

### Arguments
- **databaseId** (string) - Required - The ID of the Notion database.
- **readLimit** (number) - Optional - The maximum number of pages to read.

### Environment Variables Required
- NOTION_API_KEY
```

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

### webSearch

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Search the web or podcasts. Runs a search query through Exa, Tavily, ExaCode (code/docs), or Podscan (podcasts). Returns URL, title, and relevant Markdown snippets — without ingesting.

```APIDOC
## webSearch

### Description
Search the web or podcasts. Runs a search query through Exa, Tavily, ExaCode (code/docs), or Podscan (podcasts). Returns URL, title, and relevant Markdown snippets — without ingesting.

### Method
Not applicable (Tool call)

### Endpoint
Not applicable (Tool call)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
// Web search
{
  "tool": "webSearch",
  "arguments": {
    "query": "site:github.com graphlit MCP server",
    "searchService": "Exa",
    "limit": 5
  }
}

// Podcast search
{
  "tool": "webSearch",
  "arguments": {
    "query": "Andrej Karpathy neural networks",
    "searchService": "Podscan",
    "limit": 10
  }
}
```

### Response
#### Success Response (200)
Returns search results including URL, title, and relevant Markdown snippets. Specific schema not detailed in source.
```

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

### Text-to-Speech

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Converts text to MP3 audio using ElevenLabs and stores it in the knowledge base.

```APIDOC
## publishAudio

### Description
Converts text to MP3 audio using ElevenLabs and stores the result in the knowledge base. Retrieve the content resource to get the downloadable audio URL.

### Arguments
- **name** (string) - Required - The name for the audio content.
- **text** (string) - Required - The text to convert to audio.
- **textType** (string) - Required - The type of the text (e.g., 'Markdown').
- **voice** (string) - Required - The ID of the voice to use for synthesis.

### Response Example
```json
[ { "id": "content-guid" } ]
```
```

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

### configureProject Tool

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Configures the Graphlit project's default ingestion workflow and conversation specification. Use this tool to set up document preparation, entity extraction, LLM models, and search parameters.

```json
{
  "tool": "configureProject",
  "arguments": {
    "modelServiceType": "ANTHROPIC",          // ANTHROPIC | OPEN_AI | GOOGLE
    "configureConversationSpecification": true, // enable Claude 3.7 Sonnet + hybrid search + Cohere reranking
    "configurePreparationSpecification": true,  // enable vision-LLM document prep (extended thinking)
    "configureExtractionSpecification": true    // enable entity extraction into knowledge graph
  }
}
```

```json
{ "id": "3f2a1b4c-0000-0000-0000-000000000001" }
```

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

### Access Specification Metadata

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use `read_resource` with `specifications://` to list available specifications or retrieve the full JSON configuration for a specific LLM specification.

```typescript
read_resource("specifications://")
```

```typescript
read_resource("specifications://spec-guid")
// Full JSON of the LLM specification (model, temperature, strategy, etc.)
```

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

### ingestJiraIssues

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests issues from a Jira project. Requires Jira credentials.

```APIDOC
## ingestJiraIssues

### Description
Ingests issues from a Jira project.

### Arguments
- **url** (string) - Required - The URL of the Jira instance.
- **projectName** (string) - Required - The name of the Jira project.
- **readLimit** (number) - Optional - The maximum number of issues to read.

### Environment Variables Required
- JIRA_EMAIL, JIRA_TOKEN
```

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

### ingestSharePointFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from SharePoint. Requires SharePoint credentials.

```APIDOC
## ingestSharePointFiles

### Description
Ingests files from SharePoint.

### Arguments
- **libraryId** (string) - Required - The ID of the SharePoint library.
- **folderId** (string) - Required - The ID of the SharePoint folder.
- **readLimit** (number) - Optional - The maximum number of files to read.

### Environment Variables Required
- SHAREPOINT_ACCOUNT_NAME, SHAREPOINT_CLIENT_ID,
  SHAREPOINT_CLIENT_SECRET, SHAREPOINT_REFRESH_TOKEN
```

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

### ingestGitHubFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests source files from a GitHub repository recursively. Requires GITHUB_PERSONAL_ACCESS_TOKEN environment variable.

```APIDOC
## ingestGitHubFiles

### Description
Ingests source files from a GitHub repository recursively.

### Arguments
- **repositoryOwner** (string) - Required - The owner of the GitHub repository.
- **repositoryName** (string) - Required - The name of the GitHub repository.
- **readLimit** (number) - Optional - The maximum number of files to read.
- **recurring** (boolean) - Optional - Whether to ingest recursively.

### Environment Variables Required
- GITHUB_PERSONAL_ACCESS_TOKEN

### Response Example
```json
{
  "id": "feed-guid"
}
```
```

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

### One-shot Vector Search for Content

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Employ 'retrieveSources' for direct content lookups using hybrid search. It returns ranked content chunks with resource URIs. Suitable for single queries without maintaining conversation state.

```typescript
{
  "tool": "retrieveSources",
  "arguments": {
    "prompt": "kubernetes pod scheduling affinity rules",
    "type": "File",
    "fileType": "Document",
    "inLast": "P30D",
    "feeds": ["feed-guid-1"],
    "collections": ["collection-guid-1"]
  }
}

// Each result
{
  "id": "content-guid",
  "relevance": 0.93,
  "resourceUri": "contents://content-guid",
  "text": "...matched chunk text...",
  "mimeType": "text/markdown"
}
```

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

### Windsurf IDE mcp_config.json Configuration

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

Configure your mcp_config.json file in Windsurf IDE with these settings to specify the graphlit-mcp-server command, arguments, and environment variables.

```json
{
    "mcpServers": {
        "graphlit-mcp-server": {
            "command": "npx",
            "args": [
                "-y",
                "graphlit-mcp-server"
            ],
            "env": {
                "GRAPHLIT_ORGANIZATION_ID": "your-organization-id",
                "GRAPHLIT_ENVIRONMENT_ID": "your-environment-id",
                "GRAPHLIT_JWT_SECRET": "your-jwt-secret"
            }
        }
    }
}
```

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

### Ingest GitHub Repository Files

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests source files from a GitHub repository recursively. Requires GITHUB_PERSONAL_ACCESS_TOKEN environment variable.

```typescript
{
  "tool": "ingestGitHubFiles",
  "arguments": {
    "repositoryOwner": "openai",
    "repositoryName": "tiktoken",
    "readLimit": 500,
    "recurring": false
  }
}
// env required: GITHUB_PERSONAL_ACCESS_TOKEN
// { "id": "feed-guid" }
```

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

### Resource Enumeration Tools

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Discover resource IDs for various services like Slack, Notion, Discord, SharePoint, Dropbox, Box, Linear, and Google/Microsoft Calendars.

```APIDOC
## listSlackChannels

### Description
Lists available Slack channels.

### Arguments
- None

### Environment Variables
- SLACK_BOT_TOKEN

### Response Example
```json
[ { "name": "engineering", "id": "C01ABCDEF" }, ... ]
```
```

```APIDOC
## listNotionDatabases

### Description
Lists available Notion databases.

### Arguments
- None

### Environment Variables
- NOTION_API_KEY
```

```APIDOC
## listNotionPages

### Description
Lists Notion pages within a specific database.

### Arguments
- **databaseId** (string) - Required - The ID of the Notion database.
```

```APIDOC
## listDiscordGuilds

### Description
Lists available Discord guilds.

### Arguments
- None
```

```APIDOC
## listDiscordChannels

### Description
Lists available Discord channels within a guild.

### Arguments
- **guildId** (string) - Required - The ID of the Discord guild.
```

```APIDOC
## listSharePointLibraries

### Description
Lists available SharePoint libraries.

### Arguments
- None
```

```APIDOC
## listSharePointFolders

### Description
Lists available SharePoint folders within a library.

### Arguments
- **libraryId** (string) - Required - The ID of the SharePoint library.
```

```APIDOC
## listDropboxFolders

### Description
Lists available Dropbox folders.

### Arguments
- **folderPath** (string) - Optional - The path to the folder in Dropbox. Defaults to root.
```

```APIDOC
## listBoxFolders

### Description
Lists available Box folders.

### Arguments
- **folderId** (string) - Optional - The ID of the folder in Box. Defaults to root.
```

```APIDOC
## listLinearProjects

### Description
Lists available Linear projects.

### Arguments
- None
```

```APIDOC
## listGoogleCalendars

### Description
Lists available Google calendars.

### Arguments
- None
```

```APIDOC
## listMicrosoftCalendars

### Description
Lists available Microsoft calendars.

### Arguments
- None
```

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

### Ingest Local File

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Asynchronously ingest a local file by reading it, base64-encoding it, and sending it for processing. The MIME type is detected automatically.

```typescript
{ "tool": "ingestFile", "arguments": { "filePath": "/home/user/reports/q4-2024.pdf" } }
// { "id": "content-guid" }
```

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

### Publish Audio via ElevenLabs

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Converts provided text into an MP3 audio file using ElevenLabs' text-to-speech service. The generated audio is stored and can be retrieved via its content ID.

```typescript
{
  "tool": "publishAudio",
  "arguments": {
    "name": "Weekly Briefing - Nov 1",
    "text": "## This Week in Engineering\n\nWe shipped the new search pipeline...",
    "textType": "Markdown",
    "voice": "HqW11As4VRPkApNPkAZp"
  }
}
```

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

### Windows npx Command Execution

Source: https://github.com/graphlit/graphlit-mcp-server/blob/main/README.md

When running 'npx' on a Windows system, you might need to explicitly call it via the command prompt using this format. This ensures proper execution in the Windows environment.

```json
"command": "C:\\Windows\\System32\\cmd.exe /c npx"
```

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

### retrieveSources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Performs a hybrid (vector + keyword) search and returns ranked content chunks with resource URIs. Suitable for "one-shot" lookups where a full conversation is not needed.

```APIDOC
## retrieveSources

### Description
Performs a hybrid (vector + keyword) search and returns ranked content chunks with resource URIs. Suitable for "one-shot" lookups where a full conversation is not needed.

### Method
Not specified (Tool invocation)

### Arguments
- **prompt** (string) - Required - The search query.
- **type** (string) - Optional - The type of content to retrieve (e.g., "File").
- **fileType** (string) - Optional - The specific file type (e.g., "Document").
- **inLast** (string) - Optional - Time period for filtering results (e.g., "P30D").
- **feeds** (array of strings) - Optional - List of feed IDs to search within.
- **collections** (array of strings) - Optional - List of collection IDs to search within.

### Request Example
```json
{
  "tool": "retrieveSources",
  "arguments": {
    "prompt": "kubernetes pod scheduling affinity rules",
    "type": "File",
    "fileType": "Document",
    "inLast": "P30D",
    "feeds": ["feed-guid-1"],
    "collections": ["collection-guid-1"]
  }
}
```

### Response Example
```json
{
  "id": "content-guid",
  "relevance": 0.93,
  "resourceUri": "contents://content-guid",
  "text": "...matched chunk text...",
  "mimeType": "text/markdown"
}
```
```

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

### Ingest Project Management Issues (Linear/Jira)

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests issues from project management tools like Linear or Jira. Linear requires LINEAR_API_KEY. Jira requires JIRA_EMAIL and JIRA_TOKEN.

```typescript
// Linear
{
  "tool": "ingestLinearIssues",
  "arguments": { "projectName": "Backend Platform", "readLimit": 300 }
}
// env: LINEAR_API_KEY

// Jira
{
  "tool": "ingestJiraIssues",
  "arguments": {
    "url": "https://mycompany.atlassian.net",
    "projectName": "ENG",
    "readLimit": 200
  }
}
// env: JIRA_EMAIL, JIRA_TOKEN
```

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

### ingestDropboxFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from Dropbox. Requires Dropbox credentials.

```APIDOC
## ingestDropboxFiles

### Description
Ingests files from Dropbox.

### Arguments
- **path** (string) - Required - The path to the folder in Dropbox.
- **readLimit** (number) - Optional - The maximum number of files to read.

### Environment Variables Required
- DROPBOX_APP_KEY, DROPBOX_APP_SECRET, DROPBOX_REFRESH_TOKEN
```

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

### Generate Image via OpenAI

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Creates an image based on a text prompt using OpenAI's image generation model. The number of images to generate can be specified.

```typescript
{
  "tool": "publishImage",
  "arguments": {
    "name": "Mascot Draft",
    "prompt": "A friendly cartoon raccoon wearing a Graphlit t-shirt, digital art style",
    "count": 2
  }
}
```

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

### queryProjectUsage Tool

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Retrieves billable usage records for the Graphlit platform. This tool returns a paginated audit log of API operations, including credits consumed, token counts, and model details within a specified time window.

```json
{
  "tool": "queryProjectUsage",
  "arguments": {
    "inLast": "P1D"
  }
}
```

```json
[
  {
    "date": "2024-11-01T12:34:56Z",
    "name": "Prompt completion",
    "metric": "TOKENS",
    "credits": 0.05,
    "modelService": "Anthropic",
    "modelName": "claude-3-7-sonnet",
    "promptTokens": 1200,
    "completionTokens": 340,
    "tokens": 1540
  }
]
```

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

### ingestLinearIssues

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests issues from a Linear project. Requires LINEAR_API_KEY environment variable.

```APIDOC
## ingestLinearIssues

### Description
Ingests issues from a Linear project.

### Arguments
- **projectName** (string) - Required - The name of the Linear project.
- **readLimit** (number) - Optional - The maximum number of issues to read.

### Environment Variables Required
- LINEAR_API_KEY
```

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

### webCrawl

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Crawl a website into the knowledge base. Discovers pages via `sitemap.xml` and ingests them asynchronously. Supports one-time or recurring schedules.

```APIDOC
## webCrawl

### Description
Crawl a website into the knowledge base. Discovers pages via `sitemap.xml` and ingests them asynchronously. Supports one-time or recurring schedules.

### Method
Not applicable (Tool call)

### Endpoint
Not applicable (Tool call)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
{
  "tool": "webCrawl",
  "arguments": {
    "url": "https://docs.graphlit.dev",
    "readLimit": 200,
    "recurring": true,
    "repeatInterval": "PT1H"
  }
}
```

### Response
#### Success Response (200)
- **id** (string) - The unique identifier for the created feed.

#### Response Example
```json
{ "id": "feed-guid" }
// Then poll: { "tool": "isFeedDone", "arguments": { "id": "feed-guid" } }
```
```

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

### Ingest Notion Pages

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests pages from a Notion database. Requires NOTION_API_KEY environment variable.

```typescript
{
  "tool": "ingestNotionPages",
  "arguments": {
    "databaseId": "Engineering-Wiki-114abc10cb38487e91ec906fc6c6f350",
    "readLimit": 100
  }
}
// env required: NOTION_API_KEY
```

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

### Manage Collection Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use `read_resource` with `collections://` to list all collections or retrieve the details of a specific collection, including its associated content URIs.

```typescript
read_resource("collections://")
// [{ name: "Sprint 42 Notes", uri: "collections://collection-guid" }]
```

```typescript
read_resource("collections://collection-guid")
// { "id": "...", "name": "Sprint 42 Notes",
//   "contents": ["contents://guid-1", "contents://guid-2"] }
```

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

### Access Workflow Metadata

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Retrieve a list of all workflows or the full JSON configuration for a specific workflow using `read_resource` with the `workflows://` URI.

```typescript
read_resource("workflows://")
```

```typescript
read_resource("workflows://workflow-guid")
// Full JSON of the workflow configuration
```

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

### Initiate or Continue LLM Conversation

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use 'promptConversation' for multi-turn LLM interactions. It maintains message history and uses hybrid search for relevant content. Provide 'conversationId' to continue an existing conversation.

```typescript
// First turn
{
  "tool": "promptConversation",
  "arguments": {
    "prompt": "Summarize what our Slack #engineering channel has been discussing this week."
  }
}
// Response
{
  "id": "conv-guid-here",
  "message": "This week, the #engineering channel focused on ...",
  "citations": [
    { "index": 1, "content": { "id": "content-guid" }, "text": "..." }
  ]
}

// Follow-up turn reusing the same conversation
{
  "tool": "promptConversation",
  "arguments": {
    "prompt": "Which team members were most active?",
    "conversationId": "conv-guid-here"
  }
}
```

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

### AI Image Generation

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Generates an image from a text prompt using OpenAI's image model and stores it in the knowledge base.

```APIDOC
## publishImage

### Description
Generates an image from a text prompt using OpenAI's image model and stores it in the knowledge base.

### Arguments
- **name** (string) - Required - The name for the generated image.
- **prompt** (string) - Required - The text prompt for image generation.
- **count** (integer) - Optional - The number of images to generate.

### Response Example
```json
[ { "id": "content-guid-1" }, { "id": "content-guid-2" } ]
```
```

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

### ingestOneDriveFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from OneDrive. Requires OneDrive credentials.

```APIDOC
## ingestOneDriveFiles

### Description
Ingests files from OneDrive.

### Arguments
- **readLimit** (number) - Optional - The maximum number of files to read.

### Environment Variables Required
- ONEDRIVE_CLIENT_ID, ONEDRIVE_CLIENT_SECRET, ONEDRIVE_REFRESH_TOKEN
```

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

### ingestGoogleDriveFiles

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from Google Drive. Supports service account JSON or user OAuth. Requires Google Drive credentials. Recurring ingestion is supported.

```APIDOC
## ingestGoogleDriveFiles

### Description
Ingests files from Google Drive. Supports both service account JSON and user OAuth (client ID + refresh token).

### Arguments
- **folderId** (string) - Required - The ID of the Google Drive folder.
- **readLimit** (number) - Optional - The maximum number of files to read.
- **recurring** (boolean) - Optional - Whether to ingest files periodically.

### Environment Variables Required
- GOOGLE_DRIVE_SERVICE_ACCOUNT_JSON
  OR
- GOOGLE_DRIVE_CLIENT_ID + GOOGLE_DRIVE_CLIENT_SECRET + GOOGLE_DRIVE_REFRESH_TOKEN
```

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

### webMap

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Enumerate URLs from a sitemap. Returns all URLs found at or beneath the provided site URL via sitemap traversal, without ingesting any content.

```APIDOC
## webMap

### Description
Enumerate URLs from a sitemap. Returns all URLs found at or beneath the provided site URL via sitemap traversal, without ingesting any content.

### Method
Not applicable (Tool call)

### Endpoint
Not applicable (Tool call)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
{ "tool": "webMap", "arguments": { "url": "https://www.graphlit.com" } }
```

### Response
#### Success Response (200)
- **Array of strings** - A list of URLs found on the sitemap.

#### Response Example
```json
["https://www.graphlit.com/", "https://www.graphlit.com/blog", ...]
```
```

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

### queryProjectUsage

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Retrieves billable usage records for the Graphlit project. Returns a paginated audit log of API operations within a specified time window.

```APIDOC
## queryProjectUsage — Retrieve billable usage records

### Description
Returns a paginated audit log of all API operations in a given time window, including credits consumed, token counts, model names, and GraphQL operation details.

### Arguments
- **inLast** (string) - Required - Specifies the time window for the usage records (e.g., "P1D" for the last 24 hours).

### Request Example
```json
{
  "tool": "queryProjectUsage",
  "arguments": {
    "inLast": "P1D"
  }
}
```

### Response Example
```json
[
  {
    "date": "2024-11-01T12:34:56Z",
    "name": "Prompt completion",
    "metric": "TOKENS",
    "credits": 0.05,
    "modelService": "Anthropic",
    "modelName": "claude-3-7-sonnet",
    "promptTokens": 1200,
    "completionTokens": 340,
    "tokens": 1540
  }
]
```
```

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

### Ingest Single URL

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Asynchronously ingest a single resource from a URL, such as a web page, PDF, audio, or image. Use `isContentDone` to poll for completion.

```typescript
{ "tool": "ingestUrl", "arguments": { "url": "https://arxiv.org/pdf/2303.08774.pdf" } }
// { "id": "content-guid" }
// Poll: { "tool": "isContentDone", "arguments": { "id": "content-guid" } }
```

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

### ingestGitHubIssues

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests issues (with attachments) from a GitHub repository. Requires GITHUB_PERSONAL_ACCESS_TOKEN environment variable.

```APIDOC
## ingestGitHubIssues

### Description
Ingests issues (with attachments) from a GitHub repository.

### Arguments
- **repositoryOwner** (string) - Required - The owner of the GitHub repository.
- **repositoryName** (string) - Required - The name of the GitHub repository.
- **readLimit** (number) - Optional - The maximum number of issues to read.

### Environment Variables Required
- GITHUB_PERSONAL_ACCESS_TOKEN
```

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

### Analyze Image Content

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Uses a vision LLM to describe or analyze an image provided either by URL or by an ingested content ID. Requires a prompt specifying the analysis task.

```typescript
{ 
  "tool": "describeImageUrl",
  "arguments": {
    "prompt": "List all text visible in this diagram.",
    "url": "https://example.com/architecture-diagram.png"
  }
}
```

```typescript
{ 
  "tool": "describeImageContent",
  "arguments": {
    "id": "content-guid",
    "prompt": "What is the brand name shown in the logo?"
  }
}
```

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

### retrieveImages

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Fetches an image from a URL, creates embeddings, and returns visually similar images stored in the knowledge base. Supports optional geo-location filtering for GPS-tagged images.

```APIDOC
## retrieveImages

### Description
Fetches an image from a URL, creates embeddings, and returns visually similar images stored in the knowledge base. Supports optional geo-location filtering for GPS-tagged images.

### Method
Not specified (Tool invocation)

### Arguments
- **url** (string) - Required - The URL of the reference image.
- **location** (object) - Optional - Geo-location filter.
  - **latitude** (number) - Required - Latitude coordinate.
  - **longitude** (number) - Required - Longitude coordinate.
  - **distance** (number) - Required - Maximum distance in meters.
- **inLast** (string) - Optional - Time period for filtering results (e.g., "P7D").
- **limit** (number) - Optional - Maximum number of results to return.

### Request Example
```json
{
  "tool": "retrieveImages",
  "arguments": {
    "url": "https://example.com/reference-photo.jpg",
    "location": {
      "latitude": 47.6062,
      "longitude": -122.3321,
      "distance": 5000
    },
    "inLast": "P7D",
    "limit": 20
  }
}
```

### Response Example
```json
{
  "id": "content-guid",
  "relevance": 0.88,
  "fileName": "seattle-skyline.jpg",
  "resourceUri": "contents://content-guid",
  "uri": "https://cdn.graphlit.dev/...",
  "mimeType": "image/jpeg"
}
```
```

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

### Ingest OneDrive Files

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Ingests files from OneDrive. Requires ONEDRIVE_CLIENT_ID, ONEDRIVE_CLIENT_SECRET, and ONEDRIVE_REFRESH_TOKEN environment variables.

```typescript
{
  "tool": "ingestOneDriveFiles",
  "arguments": { "readLimit": 100 }
}
// env required: ONEDRIVE_CLIENT_ID, ONEDRIVE_CLIENT_SECRET, ONEDRIVE_REFRESH_TOKEN
```

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

### List or Search Data Connector Feeds

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use 'queryFeeds' to retrieve data connector feeds, optionally filtering by name and type.

```typescript
{
  "tool": "queryFeeds",
  "arguments": {
    "type": "Slack"
  }
}
// [{ "id": "feed-guid", "resourceUri": "feeds://feed-guid" }]
```

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

### Screenshot Web Page

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Takes a screenshot of a given URL, ingests the image, and returns a content ID for later retrieval.

```APIDOC
## screenshotPage

### Description
Takes a screenshot of a URL, ingests the image synchronously, and returns a content ID. Retrieve the content resource to get the downloadable image URL.

### Arguments
- **url** (string) - Required - The URL of the web page to screenshot.

### Response Example
```json
{ "id": "content-guid" }
```
### Usage
Then: `read resource contents://content-guid` to get imageUri
```

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

### Bulk delete resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Bulk-delete with optional filters and a limit cap.

```APIDOC
## deleteContents / deleteFeeds / deleteCollections / deleteConversations

### Description
Bulk-delete with optional filters and a limit cap.

### Method
Not applicable (Tool calls)

### Endpoint
Not applicable (Tool calls)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
```json
// Delete up to 500 audio files ingested at any time
{
  "tool": "deleteContents",
  "arguments": { "contentType": "File", "fileType": "Audio", "limit": 500 }
}

// Delete all Slack feeds (and their contents)
{ "tool": "deleteFeeds", "arguments": { "feedType": "Slack" } }
```

### Response
#### Success Response (200)
No specific response body documented for these bulk delete operations.
```

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

### promptConversation

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Initiates or continues a multi-turn LLM conversation over the entire knowledge base. It uses hybrid vector search to find relevant content and maintains full message history across calls.

```APIDOC
## promptConversation

### Description
Initiates or continues a multi-turn LLM conversation. Uses hybrid vector search to find relevant content, then completes the user prompt using the configured LLM. Maintains full message history across calls.

### Method
Not specified (Tool invocation)

### Arguments
- **prompt** (string) - Required - The user's input prompt.
- **conversationId** (string) - Optional - The ID of the conversation to continue.

### Request Example
```json
{
  "tool": "promptConversation",
  "arguments": {
    "prompt": "Summarize what our Slack #engineering channel has been discussing this week."
  }
}
```

### Response Example
```json
{
  "id": "conv-guid-here",
  "message": "This week, the #engineering channel focused on ...",
  "citations": [
    { "index": 1, "content": { "id": "content-guid" }, "text": "..." }
  ]
}
```
```

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

### Image Similarity Search

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Use 'retrieveImages' to find visually similar images in the knowledge base based on a reference URL. Supports optional geo-location filtering and date range constraints.

```typescript
{
  "tool": "retrieveImages",
  "arguments": {
    "url": "https://example.com/reference-photo.jpg",
    "location": {
      "latitude": 47.6062,
      "longitude": -122.3321,
      "distance": 5000
    },
    "inLast": "P7D",
    "limit": 20
  }
}

// Each result
{
  "id": "content-guid",
  "relevance": 0.88,
  "fileName": "seattle-skyline.jpg",
  "resourceUri": "contents://content-guid",
  "uri": "https://cdn.graphlit.dev/...",
  "mimeType": "image/jpeg"
}
```

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

### Bulk Delete Resources

Source: https://context7.com/graphlit/graphlit-mcp-server/llms.txt

Perform bulk deletions of content, feeds, collections, or conversations. Supports filtering by type and a limit cap for operations like deleting up to 500 audio files.

```typescript
// Delete up to 500 audio files ingested at any time
{
  "tool": "deleteContents",
  "arguments": { "contentType": "File", "fileType": "Audio", "limit": 500 }
}
```

```typescript
// Delete all Slack feeds (and their contents)
{ "tool": "deleteFeeds", "arguments": { "feedType": "Slack" } }
```