Automate Weekly Metrics Delivery to Slack with Paradime, OpenClaw, and Google Sheets

Stop manually pulling numbers every Friday. Wire up a Paradime + OpenClaw pipeline that reads your metrics spreadsheet, calculates week-over-week trends, and drops a formatted summary into Slack — on autopilot.

What Is Paradime?

Paradime is the all-in-one, AI-native data platform that replaces dbt Cloud™. It gives fast-moving data teams a single pane of glass to code, ship, fix, and scale data pipelines for analytics and AI. Three pillars matter for this guide:

Paradime is SOC 2 Type II certified, GDPR & CCPA compliant, and guarantees 99.9% uptime, so your credentials and pipeline state are locked down from day one.

Why does this matter here? Bolt is the scheduler that will fire our metrics script every Friday at 5 PM, and its UI-driven environment-variable vault is where sensitive keys live — no .env files scattered across laptops.

What Is OpenClaw?

OpenClaw is an open-source AI agent runtime that runs locally on your hardware and orchestrates tasks across chat apps, files, the web, and your operating system. It is not an LLM — it connects to models like Claude or GPT via API and uses skills and cron jobs to act.

Key concepts:

• Gateway — the always-on daemon that manages agent sessions, cron jobs, and channel routing.

• Skills — installable plugins (Google Sheets, Slack, GitHub, etc.) that give the agent tool access.

• Cron Jobs — built-in scheduler with at, every, and cron schedule types, plus delivery targets (Slack channel, webhook, etc.).

OpenClaw's local-first model means your Google credentials JSON and Slack tokens never leave your machine — a meaningful security win over cloud-hosted agent platforms.

Architecture Overview

Before diving into code, here is the end-to-end flow:

Figure 1: End-to-end flow — Bolt triggers OpenClaw, which reads Google Sheets, crunches numbers, and posts to Slack.

Setup: openclaw-sdk + Google Sheets API + Slack SDK

Prerequisites

1. Install and Configure OpenClaw

During onboarding, choose your preferred LLM provider (Anthropic, OpenAI, or OpenRouter) and paste in your API key. OpenClaw stores the key in ~/.openclaw/openclaw.json:

2. Install the Google Sheets Skill

Then authenticate:

This launches the OAuth flow. Once complete, tokens are stored locally — never shipped to a third-party server.

Tip: For headless / CI environments, use a GCP service-account JSON key instead of the OAuth flow. Drop the key file at a secure path and reference it via the GOOGLE_CREDENTIALS_JSON env var.

3. Set Up Your Slack App

Head to api.slack.com/apps and create a new app. Add these Bot Token Scopes:

• channels:join

• channels:read

• chat:write

• files:write

• users:read

Install the app to your workspace, copy the Bot User OAuth Token (xoxb-...), and enable Socket Mode (grab the App-Level Token xapp-...).

Wire both tokens into OpenClaw:

Or set them as environment variables:

Environment Variables

Three secrets power this pipeline. Here is where each one lives and what it does:

Storing Secrets in Paradime Bolt

Paradime's UI-driven vault is the cleanest option for production:

1. Open Settings → Workspaces → Environment Variables.

2. In the Bolt Schedules section, click Add New.

3. Enter the key name (e.g., GOOGLE_CREDENTIALS_JSON) and paste the value.

4. Hit the 💾 save icon.

Figure 2: Adding environment variables through the Paradime UI — no config files, no drift.

These variables are available to every Bolt schedule by default, but you can override them per-schedule when staging and production need different credentials.

Storing Secrets in OpenClaw

OpenClaw uses a layered precedence model (highest → lowest):

1. Process environment (parent shell / daemon)

2. .env in current working directory

3. Global .env at ~/.openclaw/.env

4. Config env block in ~/.openclaw/openclaw.json

Security note: OpenClaw never overrides an existing env var. If you set SLACK_BOT_TOKEN in your shell, the .env file will not clobber it. This protects against accidental credential leaks in shared environments.

The Script: Read Metrics, Calculate Trends, Post to Slack

Here is a complete Python script that does the heavy lifting. Save it as weekly_metrics.py in your dbt™ project repo:

How the Script Works

Figure 3: Script execution flow — authenticate, read, compute, format, post.

The script is intentionally framework-free. It uses gspread for Sheets, slack_sdk for Slack, and standard-library JSON/os — nothing else. This keeps the dependency footprint tiny and makes debugging straightforward.

Bolt Schedule: Cron Every Friday at 5 PM

Option A: UI-Driven Schedule (Recommended)

1. Navigate to Bolt → Schedules → Create Schedule.

2. Set the Schedule Type to Standard.

3. Under Commands, add:

4. Under Trigger, choose Scheduled Run and enter the cron expression:

5. Set Timezone to your team's local timezone (e.g., America/New_York).

6. Under Notifications, add your Slack channel so you get alerted if the schedule itself fails.

7. Click Save.

Pro tip: Validate your cron expression at crontab.guru before saving.

Option B: Schedules as Code (YAML)

If you prefer version-controlled schedules, add this to your paradime_schedules.yml:

Paradime scans for changes to this file every 10 minutes, or you can trigger a manual parse from the Bolt UI.

Option C: OpenClaw Cron Job

If you want OpenClaw itself to be the scheduler (instead of Bolt), create a cron job via the gateway:

When to pick which: Use Bolt when the script lives inside your dbt™ project and you want unified observability across all your dbt™ runs. Use OpenClaw cron when the agent needs to reason about the output (e.g., add commentary, flag anomalies) before posting.

Connecting dbt™ to the Picture

You might wonder: "Where does dbt™ fit in?" Great question. The metrics in your Google Sheet likely originate from dbt™ models running in your warehouse. A typical pattern looks like this:

Figure 4: dbt™ models feed the warehouse, a sync job pushes KPIs to Google Sheets, and our script delivers the summary to Slack.

If you're using Paradime's Code IDE, you can validate those models with dbt™-llm-evals for AI-generated content. A quick setup looks like this:

This gives you warehouse-native evaluation of any AI-generated content your pipelines produce — a nice complement to the metrics delivery pipeline we're building here.

Monitoring and Debugging

Bolt Run History

After your first Friday run, head to Bolt → Schedules → weekly-metrics-to-slack and click into the run. You'll see:

• Status: passed / failed / running

• Logs: full stdout/stderr from the Python script

• Duration: how long the run took

• DinoAI Debugging: if a run fails, Paradime's AI assistant can analyze the error and suggest a fix directly in the UI

OpenClaw Gateway Logs

If you're using OpenClaw as the scheduler, check the gateway logs:

Or tail the log file directly:

Slack Delivery Verification

Add a simple health check to the end of your script:

Monitoring Checklist

Figure 5: Monitoring decision tree — follow the arrows to diagnose failures fast.

Troubleshooting Common Issues

🔴 google.auth.exceptions.DefaultCredentialsError

Cause: The GOOGLE_CREDENTIALS_JSON env var is missing or contains malformed JSON.

Fix:

1. Verify the variable exists in Bolt: Settings → Workspaces → Environment Variables → Bolt Schedules.

2. Ensure the value is the entire JSON key (not a file path).

3. Test locally:

🔴 slack_sdk.errors.SlackApiError: not_in_channel

Cause: The Slack bot hasn't been added to the target channel.

Fix:

1. Open the Slack channel → click the channel name → Integrations → Add apps.

2. Find your bot and add it.

3. Alternatively, grant the channels:join scope and call client.conversations_join(channel=SLACK_CHANNEL) in your script.

🔴 gspread.exceptions.SpreadsheetNotFound

Cause: The service account email doesn't have access to the spreadsheet.

Fix:

1. Open the Google Sheet → Share → paste the service account email (found in your credentials JSON under client_email).

2. Grant Viewer access (read-only is sufficient).

🔴 ValueError: Need at least two weeks of data

Cause: Your spreadsheet has fewer than two data rows.

Fix: Ensure at least two weeks of metrics are present. The script reads the last two rows to compute deltas.

🔴 OpenClaw Cron Job Fires but No Slack Message Appears

Cause: The delivery.to channel ID is wrong, or the bot token lacks chat:write scope.

Fix:

1. Get the correct channel ID: right-click the channel in Slack → View channel details → copy the ID at the bottom.

2. Verify bot scopes at api.slack.com/apps → OAuth & Permissions.

🔴 Bolt Schedule Shows "Suspended"

Cause: The paradime_schedules.yml has suspended: true, or the schedule was manually suspended in the UI.

Fix:

1. In YAML: set suspended: false (or remove the key entirely).

2. In UI: navigate to the schedule and toggle it back to active.

Wrapping Up

Here's what we built:

1. A Python script (weekly_metrics.py) that authenticates with Google Sheets, reads the last two weeks of KPIs, computes week-over-week deltas with trend indicators, and posts a rich Block Kit summary to Slack.

2. A Paradime Bolt schedule that fires the script every Friday at 5 PM using a simple cron expression (0 17 * * 5) — with environment variables managed through a secure, UI-driven vault instead of scattered .env files.

3. An OpenClaw integration that connects Slack as a delivery channel, handles Google Sheets authentication through its skills system, and optionally serves as an alternative scheduler with built-in agent reasoning.

Figure 6: The complete stack — secrets management, scheduling, execution, and delivery.

The entire setup takes about 30 minutes. No Airflow DAGs, no Lambda functions, no Docker containers. Just a Python script, a UI-configured schedule, and secrets that stay where they belong.

What to explore next:

• Add Paradime Slack notifications to alert on dbt™ test failures alongside your metrics summary.

• Use dbt™-llm-evals to evaluate AI-generated content in your warehouse.

• Extend the OpenClaw agent to comment on anomalies (e.g., "Revenue dropped 15% — that's the steepest weekly decline this quarter") before posting.

• Explore Paradime Radar to keep your warehouse costs in check as pipeline volume grows.

Your Friday afternoon just got a lot quieter. 🎉