mAItion

Requirements

• Docker and Docker Compose
• OpenRouter or OpenAI API key (a free OpenRouter account works with the default configuration)
• S3 bucket

Setup

• Create .env.rag file by copying .env.rag.example (see https://github.com/wikiteq/rag-of-all-trades for details)
• Set OPENROUTER_API_KEY
• Set S3_ACCOUNT1_* values to match your source S3 bucket with files
• Create config.yaml out of config.yaml.example
• The default config works OK and is configured to:
• Use a single S3 bucket as data source
• Use openai/gpt-oss-20b:free model for rephrase
• Use local sentence-transformers/all-mpnet-base-v2 model for embeddings
• You can change the values if necessary, refer to https://github.com/wikiteq/rag-of-all-trades for details
• Create .env file by copying .env.openwebui.example
• Set OPENAI_API_KEY
• Optionally set OPENAI_DEFAULT_MODEL

Start the stack by running docker compose up -d. Wait until all the services become healthy. You can check health status by running docker compose ps and checking the STATUS column of the services:

docker compose ps
NAME                                      IMAGE                                      COMMAND                  SERVICE     CREATED         STATUS                   PORTS
rag-of-all-trades-openwebui-api-1         ghcr.io/wikiteq/rag-of-all-trades:latest   "sh -c 'alembic upgr…"   api         4 minutes ago   Up 4 minutes (healthy)   8000/tcp
rag-of-all-trades-openwebui-openwebui-1   ghcr.io/open-webui/open-webui:0.6.5        "/custom-entrypoint.…"   openwebui   4 minutes ago   Up 4 minutes (healthy)   0.0.0.0:3000->8080/tcp, [::]:3000->8080/tcp
rag-of-all-trades-openwebui-postgres-1    ankane/pgvector:v0.5.1                     "docker-entrypoint.s…"   postgres    4 minutes ago   Up 4 minutes (healthy)   5432/tcp
rag-of-all-trades-openwebui-redis-1       redis:7                                    "docker-entrypoint.s…"   redis       4 minutes ago   Up 4 minutes (healthy)   6379/tcp

It takes up to a minute for the OpenWebUI to fully boot on cold start.

Once all the services are booted and report healthy status visit http://localhost:3000 and login using Admin credentials. The credentials are defined in X_WEBUI_ADMIN_EMAIL and X_WEBUI_ADMIN_PASS of the .env file. The default ones are:

• username: admin@example123.com
• password: q1w2e3r4!

If you did not change the ENABLE_OPENAI_API you will also have LLM provider pre-configured with the values you have in the .env including the default chat model

Two components handle RAG service communication:

• Filter function (functions/function.py) — intercepts every user message and injects ROAT context automatically. Enabled globally via Admin Panel → Functions.
• Knowledge Base Search tool (tools/roat_retrieval.py) — a Workspace Tool that lets the LLM decide when to query ROAT. Requires a model with native function calling support. Both are automatically provisioned on first boot.

✨ Use-cases

• A single place to chat with your company knowledge that's scattered across many external systems
• A central space for looking up and refining your existing knowledge across many knowledge bases
• A tool to find secret knowledge that can not be found in the other was across your scattered data
• An entry-point into your on-premise hosted LLM models supporting evaluations and per-model settings

Quick start

Connectors configuration

The service supports multiple data sources, including multiple data sources of the same type, each with its own ingestion schedule. The connectors to enable are defined via config.yaml, and their secrets are defined in the .env.rag file.

config.yaml

sources: - type: "s3" # must be s3 name: "account1" # arbitrary name for the connector, will be stored in metadata config: endpoint: "${S3_ACCOUNT1_ENDPOINT}" # s3 endpoint access_key: "${S3_ACCOUNT1_ACCESS_KEY}" # s3 access key secret_key: "${S3_ACCOUNT1_SECRET_KEY}" # s3 secret key region: "${S3_ACCOUNT1_REGION}" # s3 region use_ssl: "${S3_ACCOUNT1_USE_SSL}" # use ssl for s3 connection, can be True or False buckets: "${S3_ACCOUNT1_BUCKETS}" # single entry or comma-separated list i.e. bucket1,bucket2 schedules: "${S3_ACCOUNT1_SCHEDULES}" # single entry or comma-separated list i.e. 3600,60

- type: "s3" name: "account2" config: ...

- type: "s3" name: "account3" config: ...

.env.rag

S3_ACCOUNT1_ENDPOINT=https://s3.amazonaws.com S3_ACCOUNT1_ACCESS_KEY=xxx S3_ACCOUNT1_SECRET_KEY=xxx S3_ACCOUNT1_REGION=us-east-1 S3_ACCOUNT1_USE_SSL=True S3_ACCOUNT1_BUCKETS=bucket1,bucket2 S3_ACCOUNT1_SCHEDULES=3600,60 ```

config.yaml

sources: - type: "mediawiki" name: "wiki1" config: host: "${MEDIAWIKI1_HOST}" path: "/w/" # optional, default /w/ scheme: "https" # optional, default https page_limit: 500 # optional, max pages per namespace (default: unlimited) namespaces: "0,1" # optional, comma-separated namespace IDs (default: content namespaces) filter_redirects: true # optional, exclude redirect pages (default: true) username: "${MEDIAWIKI1_USERNAME}" # optional, for private wikis password: "${MEDIAWIKI1_PASSWORD}" # optional, for private wikis schedules: "${MEDIAWIKI1_SCHEDULES}"

- type: "mediawiki" name: "wiki2" config: host: "${MEDIAWIKI2_HOST}" schedules: "${MEDIAWIKI2_SCHEDULES}"

.env.rag

MEDIAWIKI1_HOST=wiki.example.org MEDIAWIKI1_SCHEDULES=3600

config.yaml

sources: - type: "serpapi" name: "serp_ingestion1" config: api_key: "${SERPAPI1_KEY}" queries: "${SERPAPI1_QUERIES}" schedules: "${SERPAPI1_SCHEDULES}"

- type: "serpapi" name: "serp_ingestion2" config:

- type: "serpapi" name: "serp_ingestion3" config:

.env.rag

SERPAPI1_KEY=xxxx SERPAPI1_QUERIES=aaa SERPAPI1_SCHEDULES=3600 ```

config.yaml

sources: # URLs mode - type: "web" name: "web1" config: urls: "${WEB1_URLS}" html_to_text: true schedules: "${WEB1_SCHEDULES}"

# Sitemap mode - type: "web" name: "web2" config: sitemap_url: "${WEB2_SITEMAP_URL}" include_prefix: "${WEB2_INCLUDE_PREFIX}" html_to_text: true schedules: "${WEB2_SCHEDULES}"

.env.rag

WEB1_URLS=https://example.com/page1,https://example.com/page2 WEB1_SCHEDULES=60 WEB2_SITEMAP_URL=https://example.com/sitemap.xml WEB2_INCLUDE_PREFIX=/blog/ WEB2_SCHEDULES=60 ```

config.yaml

sources: - type: "jira" name: "jira1" config: server_url: "${JIRA1_SERVER_URL}" auth_type: "basic" # "basic" (email + API token) or "token" (PAT) email: "${JIRA1_EMAIL}" # required when auth_type is "basic" api_token: "${JIRA1_API_TOKEN}" jql: "${JIRA1_JQL}" # JQL query to select issues max_results: 50 load_comments: false max_comments: 10 schedules: "${JIRA1_SCHEDULES}"

.env.rag

JIRA1_SERVER_URL=https://your-org.atlassian.net JIRA1_EMAIL=your-email@example.com JIRA1_API_TOKEN=your-api-token JIRA1_JQL=project = MYPROJECT ORDER BY updated DESC JIRA1_SCHEDULES=3600 ```

config.yaml

sources: - type: "pipedrive" name: "pipedrive1" config: api_token: "${PIPEDRIVE1_API_TOKEN}" load_types: # optional, default: all entity types - deals - notes - persons - mails max_items: 500 # optional, per-entity limit (default: unlimited) max_retries: 3 # optional, retry attempts on failure (default: 3) filter_mail_folders: # optional, default: [inbox] - inbox - sent schedules: "${PIPEDRIVE1_SCHEDULES}"

- type: "pipedrive" name: "pipedrive2" config: api_token: "${PIPEDRIVE2_API_TOKEN}" schedules: "${PIPEDRIVE2_SCHEDULES}"

.env.rag

PIPEDRIVE1_API_TOKEN=your-pipedrive-api-token PIPEDRIVE1_SCHEDULES=3600 PIPEDRIVE2_API_TOKEN=your-second-pipedrive-api-token PIPEDRIVE2_SCHEDULES=3600 ```

config.yaml

embedding: provider: local # you can use any embedding model supported by HuggingFace model_config: sentence-transformers/all-MiniLM-L6-v2 embedding_dim: 384

inference: provider: None model_config: None ```

config.yaml

embedding: provider: openrouter model_config: text-embedding-3-small embedding_dim: 1536

inference: provider: None model_config: None ```

You must set OPENROUTER_API_KEY and OPENROUTER_API_BASE in the .env.rag file.

config.yaml

embedding: provider: openrouter model_config: text-embedding-3-small embedding_dim: 1536

inference: provider: openrouter model_config: gpt-4o ```

You must set OPENROUTER_API_KEY and OPENROUTER_API_BASE in the .env.rag file.

Reference of the `config.yaml`

The config.yaml file contains the main configuration of the service.

Environment variables (${...}) in the config file are evaluated at runtime.

```yaml sources: # holds the list of sources to ingest from (Connectors)

- type: # type of the connector (s3, mediawiki, serpapi) name: # arbitrary name for the connector, will be stored in metadata config: # connector specific configuration schedules: "${S3_ACCOUNT1_SCHEDULES}"

configures models and dimensions for embeddings

embedding: provider: openrouter # openrouter/openai or local for local HuggingFace embeddings model_config: text-embedding-3-small # model to use embedding_dim: 1536 # dimensions (check with the model docs)

configures the LLM provider and model

inference: provider: openrouter # openrouter/openai model_config: gpt-4o # model to use

vector store configuration

vector_store: table_name: embeddings hybrid_search: true # whether to use hybrid search or not chunk_size: 512 # chunk size for vector indexing chunk_overlap: 50 # overlap between chunks # hnsw indexes settings hnsw: hnsw_m: 16 # number of neighbors hnsw_ef_construction: 64 # ef construction parameter for HNSW hnsw_ef_search: 40 # ef search parameter for HNSW hnsw_dist_method: vector_cosine_ops # distance metric for HNSW ```

SerpAPI Connector

The SerpAPI connector ingests documents from Google Search results and converts them to Markdown format. The connector has the following configuration options:

```yaml

Troubleshooting