Skip to content

Filter Plugins

Filter plugins process and modify messages before they are sent to the LLM or after responses are generated.

What are Filters?

Filters act as middleware in the message pipeline:

  • Inlet: Process user messages before they reach the LLM
  • Outlet: Process LLM responses before they're displayed
  • :material-stream: Stream: Process streaming responses in real-time

Available Filter Plugins

  • Async Context Compression

    Reduces token consumption in long conversations through intelligent summarization while maintaining coherence.

    Version: 1.2.2

    Documentation

  • Context Enhancement

    Enhances chat context with additional information for better responses.

    Version: 0.2

    Documentation

  • Folder Memory

    Automatically extracts consistent "Project Rules" from ongoing conversations within a folder and injects them back into the folder's system prompt.

    Version: 0.1.0

    Documentation

  • Markdown Normalizer

    Fixes common Markdown formatting issues in LLM outputs, including Mermaid syntax, code blocks, and LaTeX formulas.

    Version: 1.2.4

    Documentation

  • Multi-Model Context Merger

    Automatically merges context from multiple model responses in the previous turn, enabling collaborative answers.

    Version: 0.1.0

    Documentation

  • Web Gemini Multimodal Filter

    A powerful filter that provides multimodal capabilities (PDF, Office, Images, Audio, Video) to any model in OpenWebUI.

    Version: 0.3.2

    Documentation

How Filters Work

graph LR
    A[User Message] --> B[Inlet Filter]
    B --> C[LLM]
    C --> D[Outlet Filter]
    D --> E[Display to User]

Inlet Processing

The inlet method processes messages before they reach the LLM:

async def inlet(self, body: dict, __metadata__: dict) -> dict:
    # Modify the request before sending to LLM
    messages = body.get("messages", [])
    # Add context, modify prompts, etc.
    return body

Outlet Processing

The outlet method processes responses after they're generated:

async def outlet(self, body: dict, __metadata__: dict) -> dict:
    # Modify the response before displaying
    messages = body.get("messages", [])
    # Format output, add citations, etc.
    return body

Quick Installation

  1. Download the desired filter .py file
  2. Navigate to Admin Panel → Settings → Functions
  3. Upload the file and configure settings
  4. Enable the filter in chat settings or globally

Development Template

"""
title: My Custom Filter
author: Your Name
version: 1.0.0
description: Description of your filter plugin
"""

from pydantic import BaseModel, Field
from typing import Optional

class Filter:
    class Valves(BaseModel):
        priority: int = Field(
            default=0,
            description="Filter priority (lower = earlier execution)"
        )
        enabled: bool = Field(
            default=True,
            description="Enable/disable this filter"
        )

    def __init__(self):
        self.valves = self.Valves()

    async def inlet(
        self,
        body: dict,
        __user__: Optional[dict] = None,
        __metadata__: Optional[dict] = None
    ) -> dict:
        """Process messages before sending to LLM."""
        if not self.valves.enabled:
            return body

        # Your inlet logic here
        messages = body.get("messages", [])

        return body

    async def outlet(
        self,
        body: dict,
        __user__: Optional[dict] = None,
        __metadata__: Optional[dict] = None
    ) -> dict:
        """Process responses before displaying."""
        if not self.valves.enabled:
            return body

        # Your outlet logic here

        return body

For more details, check our Plugin Development Guide.