GitHubVerifyWise Home

Integrations

Slack Integration

Overview

VerifyWise integrates with Slack workspaces to deliver real-time notifications and alerts directly to your team's channels. The integration supports multiple workspaces, configurable notification routing, and both manual and scheduled notifications.

Our Slack integration uses OAuth 2.0 for secure, scoped access, then routes five notification types to the right channels with full multi-channel support. You get real-time alerts for system events, plus scheduled daily reminders powered by a job queue. It automatically detects and handles invalid or archived channels, and keeps everything locked down with encrypted credential storage and TLS.

Quick start

Prerequisites

Before setting up Slack integration, ensure you have:

  1. Slack Workspace: Admin access to create and configure apps

  2. VerifyWise Account: User account with appropriate permissions

  3. Environment Variables: Required configuration in .env

Basic setup

Configure environment variables: Add these to Servers/.env

# Encryption Settings
ENCRYPTION_ALGORITHM="aes-256-cbc"
ENCRYPTION_PASSWORD="aaaa bbbb cccc dddd"

# Slack OAuth Configuration
# Get these information from Slack App Settings
SLACK_URL=https://slack.com/oauth/v2/authorize
SLACK_API_URL=https://slack.com/api/oauth.v2.access
SLACK_CLIENT_ID=1234567890.1234567890
SLACK_CLIENT_SECRET=abcdef1234567890abcdef1234567890
SLACK_USER_OAUTH_TOKEN=xoxp-1234567890987654322345678
SLACK_BOT_TOKEN=xoxb-234567wsdfgbn76543345665445678954

# Redis (for job queue)
REDIS_HOST=127.0.0.1
REDIS_PORT=6379

And then add these to Clients/.env

# Slack Settings
VITE_SLACK_CLIENT_ID=1234567890.1234567890
VITE_SLACK_URL=https://slack.com/oauth/v2/authorize
VITE_IS_SLACK_VISIBLE=true

Now, create Slack App at https://api.slack.com/apps, configure OAuth scopes (detailed below) and add to workspace from VerifyWise settings.

Quick integration steps

  1. Navigate to Settings → Slack in VerifyWise

  2. Click "Add to Slack" button

  3. Select your workspace and authorize the app

  4. Choose a channel for notifications

  5. Configure notification routing (optional)

  6. Send a test message to verify connectivity

Features

Supported notification types

VerifyWise routes five types of notifications to Slack channels:

Notification Type
Description
Example Triggers

Membership and Roles

User additions, removals, role assignments, and team management changes

New user joined, Role changed to Admin

Projects and Organizations

Project creation, modifications, and organization linkages

New project created, Project linked to org

Policy Reminders and Status

Automated policy review reminders and compliance status updates

Policy review due in 7 days, Compliance updated

Evidence and Task Alerts

Evidence uploads, reviews, task assignments, and completions

Evidence uploaded, Task assigned to user

Control or Policy Changes

Updates to key controls or policies that affect compliance and risk posture

Control updated, Policy modified

Channel configuration

  • Multiple Channels: Connect multiple Slack channels from different workspaces

  • Flexible Routing: Send different notification types to different channels

  • Multi-Channel Routing: Route a single notification type to multiple channels

  • Channel Verification: Test connectivity with "Send Test" button

  • Auto-Deactivation: Automatically disable webhooks for archived/deleted channels

Scheduled notifications

VerifyWise sends automated notifications based on configured schedules:

  • Policy Due Soon Reminders: Daily at 9:00 AM UTC

  • Compliance Status Updates: Based on policy review dates

  • Background Processing: Queue-based system ensures reliable delivery

User guide

Adding a Slack integration

Step 1: Navigate to Settings

  • Go to Settings in VerifyWise

  • Click on Slack tab

Step 2: Authorize workspace

  • Click the "Add to Slack" button

  • Select your Slack workspace from the dropdown

  • Choose the channel where notifications will be sent

  • Click "Allow" to authorize VerifyWise

Step 3: Verify integration

  • Integration appears in the table with:

    • Team name (workspace)

    • Channel name

    • Creation date

    • Active status

  • Click "Send Test" to verify connectivity

Step 4: Configure notification routing (optional)

  • Click "Configure" button

  • For each notification type, select destination channel(s)

  • Click "Send Test" to verify routing

  • Click "Save Changes"

Notes:

  • A notification type can route to multiple channels

  • A channel can receive multiple notification types

  • Leaving a notification type empty means no routing for that type

  • Changes take effect immediately

Managing integrations

View all integrations

  • Table displays all connected workspaces and channels

  • Columns: Team Name, Channel, Creation Date, Active status, Actions

Deactivate integration

  • Integrations automatically deactivate if:

    • Channel is archived in Slack, is deleted or the bot is removed from channel

    • Status changes to "No" in Active column

    • No notifications sent to inactive integrations

Remove integration

  • Currently requires manual deletion from workspace

  • Contact system administrator

Slack app setup

Creating a Slack app

  1. Go to https://api.slack.com/apps

  2. Click "Create New App"

  3. Choose "From scratch"

  4. Enter app name: VerifyWise (or your preferred name)

  5. Select workspace: Choose development workspace

  6. Click "Create App"

Configuring OAuth & permissions

Step 1: Add redirect URLs

  1. Navigate to OAuth & Permissions

  2. Scroll to Redirect URLs

  3. Add development URL: http://localhost:3000/setting/?activeTab=slack

  4. Add production URL: https://your-domain.com/setting/?activeTab=slack

  5. Click "Save URLs"

Step 2: Configure bot token scopes

  1. Scroll to Scopes section

  2. Under Bot Token Scopes, click "Add an OAuth Scope"

  3. Add these scopes:

    • channels:read

    • channels:manage

    • chat:write

    • chat:write.public

    • groups:write

    • groups:read

    • im:read

    • mpim:read

Step 3: Configure user token scopes

  1. Under User Token Scopes, click "Add an OAuth Scope"

  2. Add these scopes:

    • channels:read

    • channels:write.invites

    • groups:read

    • groups:write.invites

    • channels:write

    • chat:write

    • im:read

    • mpim:read

Installing to workspace

  1. Navigate to Install App in left sidebar

  2. Click "Install to Workspace"

  3. Review permissions and click "Allow"

  4. Copy Bot User OAuth Token (starts with xoxb-)

  5. Copy Signing Secret from Basic Information

Retrieving credentials

Client ID & Secret:

  1. Navigate to Basic Information

  2. Scroll to App Credentials

  3. Copy Client ID → Add to SLACK_CLIENT_ID

  4. Click Show next to Client Secret → Copy → Add to SLACK_CLIENT_SECRET

OAuth URLs:

  • Authorization URL: https://slack.com/oauth/v2/authorize → SLACK_URL

  • Token URL: https://slack.com/api/oauth.v2.access → SLACK_API_URL

Enabling incoming webhooks

  1. Navigate to Incoming Webhooks

  2. Toggle Activate Incoming Webhooks to On

  3. This allows VerifyWise to send formatted messages

Enabling bots

  1. Navigate to App Home

  2. Under Your App's Presence in Slack:

    • Display Name: VerifyWise (or your preference)

    • Default Username: @verifywise

  3. Enable Always Show My Bot as Online

Environment configuration

Required environment variables

# Slack OAuth Configuration
SLACK_URL=https://slack.com/oauth/v2/authorize
SLACK_API_URL=https://slack.com/api/oauth.v2.access
SLACK_CLIENT_ID=1234567890.1234567890
SLACK_CLIENT_SECRET=abcdef1234567890abcdef1234567890

# Frontend Configuration
FRONTEND_URL=https://verifywise.com

# Database (PostgreSQL)
DATABASE_URL=postgresql://user:password@localhost:5432/verifywise

# Redis (for job queue)
REDIS_HOST=127.0.0.1
REDIS_PORT=6379

# Encryption (for sensitive data)
ENCRYPTION_KEY=<32-byte-hex-string>

Development vs production

Development:

FRONTEND_URL=http://localhost:3000
SLACK_CLIENT_ID=<dev-app-client-id>
SLACK_CLIENT_SECRET=<dev-app-client-secret>

Production:

FRONTEND_URL=https://verifywise.com
SLACK_CLIENT_ID=<prod-app-client-id>
SLACK_CLIENT_SECRET=<prod-app-client-secret>

Note: Use separate Slack apps for development and production environments, if required.

MLflow Integration

Introduction

The VerifyWise MLflow integration keeps your centralized model inventory in sync with the models that data science teams register inside MLflow. Once connected, VerifyWise periodically ingests model metadata, lifecycle stages, training evidence, and experiment context so governance and compliance teams can review the same source of truth that engineers see in MLflow.

This guide explains how to configure the integration on the Integrations page and how to work with the synchronized data that appears in Model Inventory → MLFlow data.

Why it matters

Connecting VerifyWise to MLflow closes the gap between experimentation and oversight by eliminating manual exports, ensuring every production-relevant version is captured with run metrics, and surfacing governance gaps when a model lacks bias, security, or privacy evidence.

Scheduled syncs also create a defensible audit trail that shows when VerifyWise last contacted MLflow and whether any errors occurred, reducing the operational risk of silent failures or idle BullMQ queues.

Prerequisites

Before configuring the integration you need:

  • VerifyWise admin access. Only admins can open /integrations, edit the MLflow card, or start connection tests.

  • MLflow tracking server details. Collect the base URL (e.g., https://mlflow.company.com:5000), authentication method, and any credentials. The integration is read-only, but the MLflow account must list/describe models and runs.

  • Background worker availability. Ensure the VerifyWise BullMQ worker process is running in the same environment as the API so scheduled sync jobs execute after you complete a successful connection test.

VerifyWise supports one MLflow configuration per organization. Each tenant stores its own credentials, test history, and sync state so multi-tenant deployments stay isolated.

Configuring MLflow in the Integrations page

  1. Navigate to Integrations → MLflow (or use the shortcut card on the dashboard). Non-admin users will not see this page.

  2. Fill in the Tracking server URL and Request timeout. The timeout controls how long VerifyWise waits before declaring the MLflow API unreachable; increase it for slower on-prem clusters.

  3. Choose an Authentication method:

    • None – use when the server allows anonymous read access.

    • Basic auth – prompts for username and password fields directly under the selector.

    • API token – prompts for a personal access token or service token field.

  4. If your authentication method requires extras (passwords or tokens), enter them in the newly displayed inputs. Everything you type is stored encrypted and never logged.

  5. Click Test connection. Successful tests immediately update the Connection status chip, record the timestamp in the “Last successful test” card, and enable scheduled syncs. Failures surface the error, keep syncs disabled, and allow you to retry without refreshing the page.

  6. After a successful test, click Save configuration to persist the settings. VerifyWise enforces a “test-before-save” rule to prevent stale credentials from enabling the sync scheduler.

Status cards on the Integrations page

The MLflow card at the top of the page mirrors the compact cards used inside Model Inventory so you get the same at-a-glance health signals:

  • Connection status shows Connected, Action required, or Error, using VerifyWise chip colors.

  • Last error records the most recent failure reason (timeout, authorization, DNS, etc.). It clears itself after the next successful test or sync.

  • Last successful test and Last scheduled sync display their timestamps in muted text, with icons in the top-right corner similar to the Model Overview cards.

  • Upcoming sync indicates when the hourly (or temporarily accelerated) BullMQ job will run next.

Use the Re-run sync button to enqueue an immediate job without waiting for the next cron window. A slim spacer keeps the button close to the cards so the action remains visible.

Scheduled sync behavior

VerifyWise schedules an hourly BullMQ job per organization after the first successful test. (During debugging you can temporarily switch to minute-level intervals, but production defaults back to hourly.). Jobs run inside the dedicated worker process; on production deployments, ensure npm run worker (or its PM2/systemd equivalent) is started alongside the API so queues never sit idle.

The job pulls registered models, versions, and their linked runs via the MLflow REST API, then upserts them into VerifyWise’s mlflow_model_records table with an organization-specific unique key. Duplicate inserts are suppressed, and only material changes update the record timestamps.

Every attempt stamps the “Last scheduled sync” card and writes to the sync history table, even when no models changed, which helps audit teams prove monitoring coverage.

Working with MLFlow data

Once a sync runs, your model governance users can navigate to Model Inventory → MLFlow data to inspect the ingested catalog.

Summary cards and refresh control

Four header cards (“Models”, “Active”, “Staging”, “Experiments”) appear at the top of the tab. They re-use VerifyWise’s dashboard card component, including muted icons and hover states, and summarize what is currently stored for your organization. A Refresh button above the cards re-triggers the API call that fetches data from the VerifyWise backend (not MLflow directly), so it is safe to click multiple times without spamming the remote MLflow server.

Data table

The table lists each synchronized model version with columns for Model Name, Version, Lifecycle Status, Created date, Last Updated date, Description, and Actions. Pagination controls—identical to the Model Risks table—live directly beneath the grid and show “Rows per page”, an inline selector, page counts, and navigation arrows. Click any row to open the detail drawer, which includes:

  • High-level metadata (version, lifecycle stage, run ID, creation time).

  • Description text, tags, metrics, and parameters pulled from the MLflow run.

  • Experiment information (ID, name, artifact location) so you can trace lineage.

Use the action button in the last column to inspect a model in more depth without leaving VerifyWise.

Operational guidance and troubleshooting

  • Connection succeeds but no data appears Confirm the worker job is running and that the MLflow service account has permission to list registered models. The “Last scheduled sync” card should advance every hour; if it does not, expand the API logs for BullMQ worker errors.

  • Seeing Unable to reach the MLflow backend warnings Increase the Request timeout field, verify network access between VerifyWise and MLflow, and re-test the connection. The error card will capture the latest failure reason so you can share it with infrastructure teams.

  • Manual refresh vs. re-run sync: The Refresh button on the MLFlow data tab only re-queries VerifyWise’s database; use Re-run sync on the Integrations page when you need VerifyWise to pull fresh data from MLflow immediately.

PreviousModel inventoryNextIncident management

Last updated

Was this helpful?