30 Claude Prompts That Build Workflow Automations

You are an operations automation engineer who builds no-code finance workflows. <context> I want an automation that files every new invoice into my accounting tool and alerts the team, delivered as a precise, ready-to-build step-by-step scenario I can recreate in Zapier or Make without guessing. </context> <inputs> - Trigger source: [E.G. STRIPE INVOICE PAID, PAYPAL, INVOICE APP] - Accounting tool: [QUICKBOOKS / XERO / FRESHBOOKS] - Fields to map: [AMOUNT, CUSTOMER, DATE, TAX, INVOICE #] - Where to alert: [SLACK CHANNEL / EMAIL] - Edge cases to handle: [REFUNDS, DUPLICATES, CURRENCY] </inputs> <task> Build the complete scenario: the trigger, each action step in order, the exact field-to-field mapping table, a filter or path for the edge cases, and the final Slack or email notification with a sample message. Number every step and name the app plus action for each. </task> <constraints> - Real, buildable steps using apps that exist; no invented modules. - Include a field-mapping table and note any formatter or date-parse steps needed. - Handle duplicates and refunds explicitly so it never double-posts. </constraints> <format> Return the automation as a numbered step-by-step spec plus a mapping table, then a short note on how to test it with one sample invoice before going live. </format>

You are a RevOps automation specialist. <context> I need an automation that captures every inbound lead, enriches it, and assigns it fairly to a sales rep, delivered as a precise step-by-step scenario I can build in Make or Zapier. </context> <inputs> - Lead source: [WEBSITE FORM / TYPEFORM / FACEBOOK LEAD ADS] - Enrichment step (optional): [CLEARBIT / APOLLO / NONE] - CRM: [HUBSPOT / PIPEDRIVE / SALESFORCE] - Reps to rotate through: [LIST NAMES OR EMAILS] - Routing rule: [PURE ROUND-ROBIN / BY REGION / BY DEAL SIZE] - Notify rep via: [SLACK DM / EMAIL] </inputs> <task> Build the scenario end to end: trigger, optional enrichment, a data store or counter that tracks whose turn it is, the assignment logic with the exact routing rule, create or update the CRM record with the owner set, and notify the assigned rep. Number each step and name the app plus action. </task> <constraints> - Show how to persist the round-robin counter (data store or sheet) so it survives runs. - Handle a rep being out-of-office (skip rule) and a missing email (fallback owner). - No invented modules; only real apps. </constraints> <format> Return the numbered scenario plus the routing-logic pseudocode, then a short note on how to reset the rotation and add a new rep later. </format>

You are a backend engineer who writes reliable scheduled jobs. <context> I need a runnable cron job that finds unsigned contracts and sends reminder emails, delivered as one self-contained, commented Node.js script I can drop into a scheduler. </context> <inputs> - Where contracts live: [DATABASE TABLE / AIRTABLE / API ENDPOINT] - The 'unsigned' condition: [STATUS FIELD PLUS AGE, E.G. status=sent AND >3 days] - Email provider: [RESEND / SENDGRID / SMTP] - Reminder cadence: [E.G. DAY 3, 7, 14 THEN STOP] - Env vars available: [API KEYS / DB URL] </inputs> <task> Write a Node script that connects to the source, queries unsigned contracts matching the age rule, decides which reminder in the cadence each one is due for, sends the email via the provider, records that the reminder was sent to avoid repeats, and logs a summary. Include a package.json snippet and the crontab line to run it daily. </task> <constraints> - One self-contained script, async/await, try/catch around network calls, no secrets hardcoded (read from env). - Idempotent: never send the same reminder twice for the same cadence step. - Comment each block. </constraints> <format> Return the full script in a code block, then the crontab line and a note on how to dry-run it without sending real emails. </format>

You are an operations automation consultant who designs customer onboarding flows. <context> I want a new paying customer to trigger a full onboarding sequence automatically, delivered as a precise step-by-step scenario I can build in Zapier or Make. </context> <inputs> - Signal that fires it: [STRIPE PAYMENT / CRM STAGE = WON / SIGNUP] - Tools in my stack: [CRM, EMAIL TOOL, TASK/PM TOOL, SLACK] - Onboarding steps to automate: [WELCOME EMAIL, CREATE ACCOUNT, KICKOFF TASK, SCHEDULE CHECK-IN] - Owner to assign internal tasks to: [NAME/TEAM] - Timing: [IMMEDIATE VS DELAYED STEPS] </inputs> <task> Build the scenario: trigger, send the welcome email, create the customer record or project, generate the internal onboarding checklist as tasks with due dates, post a 'new customer' message to Slack, and schedule a delayed check-in step (e.g. day 7). Number each step, name the app plus action, and mark which steps are delayed. </task> <constraints> - Real apps only; show delay modules explicitly with their wait times. - Include the exact task list with relative due dates. - Note where a human approval gate belongs, if any. </constraints> <format> Return the numbered scenario plus the onboarding task checklist, then a short note on how to branch it by plan tier. </format>

You are a data engineer who builds internal reporting bots. <context> I want an automated morning digest that pulls yesterday's sales numbers and posts them to Slack, delivered as one self-contained, commented script I can schedule. </context> <inputs> - Data sources: [STRIPE / CRM API / DATABASE] - Metrics to include: [REVENUE, NEW CUSTOMERS, TOP DEAL, CHURN] - Comparison: [VS PRIOR DAY / VS SAME DAY LAST WEEK] - Slack destination: [WEBHOOK URL / CHANNEL] - Language/runtime: [PYTHON / NODE] </inputs> <task> Write a script that fetches yesterday's data from each source, computes the metrics and the comparison deltas, formats a clean Slack message with Block Kit (headline number, metric rows with up/down arrows, and the top deal), posts it via the incoming webhook, and exits cleanly. Include the crontab line to run at 8am. </task> <constraints> - One self-contained script; read tokens from env; try/catch on every API call. - Format numbers as currency and show percent change with direction. - Fail loudly to a fallback channel if a source is down. </constraints> <format> Return the full script in a code block, then the crontab entry and a note on how to add a new metric row. </format>

You are a marketing automation engineer. <context> I want to write a post once and have it reformatted and published to several social platforms automatically, delivered as a precise step-by-step scenario for Make or Zapier. </context> <inputs> - Source of the post: [AIRTABLE ROW / GOOGLE SHEET / NOTION] - Platforms to publish to: [LINKEDIN, X, INSTAGRAM, FACEBOOK PAGE] - Per-platform tweaks: [CHAR LIMITS, HASHTAGS, IMAGE SIZES] - Trigger: [STATUS = READY / SCHEDULED TIME] - Where to log results: [SHEET / SLACK] </inputs> <task> Build the scenario: trigger on a ready post, branch into one path per platform that reformats the text to that platform's rules (truncate for X, add hashtags for Instagram, and so on), publish via each platform's app or action, then log the post URL and status back to the source and alert on any failure. Number each step and name the app plus action. </task> <constraints> - Real apps only; show the text-transform step per platform explicitly. - Handle one platform failing without blocking the others (independent paths). - Mark the row as posted so it never re-publishes. </constraints> <format> Return the numbered scenario with the per-platform formatting rules, then a note on how to add image handling and a per-platform schedule offset. </format>

You are a backend engineer who wires marketing tools together. <context> When someone subscribes on my site I want them synced to my email tool, tagged, and sent a welcome, delivered as one self-contained, commented webhook handler I can deploy as a serverless function. </context> <inputs> - Signup source and payload shape: [FORM POST WITH email, name, source] - Email/ESP: [MAILERLITE / BEEHIIV / CONVERTKIT / RESEND] - Tag or list to add to: [LIST ID / TAG] - Welcome email: [SEND VIA ESP AUTOMATION / SEND DIRECTLY] - Runtime: [NODE SERVERLESS / EXPRESS] </inputs> <task> Write a handler that validates the incoming payload, upserts the subscriber into the ESP with the right list or tag, triggers or sends the welcome email, handles the already-subscribed case gracefully, and returns proper HTTP status codes. Include input validation and a signature or secret check. </task> <constraints> - One self-contained handler; read keys from env; validate email format. - Return 200 on success, 4xx on bad input, 5xx on ESP failure; never crash on malformed JSON. - Idempotent: re-sending the same signup must not duplicate or re-welcome. </constraints> <format> Return the full handler in a code block, then a note on how to test it with curl and how to point your form at it. </format>

You are an e-commerce lifecycle automation specialist. <context> I want an automated sequence that wins back shoppers who abandon their cart, delivered as a precise step-by-step scenario for Klaviyo, Make, or Zapier. </context> <inputs> - Store platform: [SHOPIFY / WOOCOMMERCE / STRIPE CHECKOUT] - Abandon trigger and wait: [CHECKOUT STARTED, NO PURCHASE IN 1 HOUR] - Channels: [EMAIL / SMS / BOTH] - Incentive rules: [NO DISCOUNT FIRST, THEN 10% ON LAST STEP] - Stop condition: [PURCHASE COMPLETED] </inputs> <task> Build the sequence: trigger on abandonment, wait, send message 1 (reminder), wait, send message 2 (social proof or objection), wait, send message 3 (incentive), each with an exit condition if they buy. Provide the timing, the copy for each message, and the branch that halts the flow on purchase. Number each step. </task> <constraints> - Real apps or modules only; show every wait and the purchase-exit check explicitly. - Include ready-to-use subject lines and body copy for each message. - Never send a discount before the final step; respect the stop condition. </constraints> <format> Return the numbered flow with timing and the message copy, then a note on how to measure recovered revenue. </format>

You are a marketing-ops engineer. <context> I want a tool that generates consistent UTM-tagged campaign URLs and logs every one it makes, delivered as one self-contained, commented script. </context> <inputs> - Base URLs I tag: [LIST OR SINGLE DOMAIN] - UTM convention: [source, medium, campaign, content, term RULES] - Naming rules: [LOWERCASE, HYPHENS, NO SPACES] - Where to log: [GOOGLE SHEET / CSV FILE] - Runtime: [PYTHON / NODE] </inputs> <task> Write a script that takes campaign inputs (either CLI args or a small config array), validates them against the naming rules, builds correctly encoded UTM URLs, appends each URL with its parameters and a timestamp to the log, and prints the finished links. Reject inputs that break the convention with a clear error. </task> <constraints> - One self-contained script; URL-encode values; enforce lowercase-hyphen naming. - Never produce a malformed or double-question-mark URL; dedupe against existing log rows. - Comment each function. </constraints> <format> Return the full script in a code block, then a note on how to run it for a batch of links from a CSV. </format>

You are a demand-gen automation consultant. <context> When someone requests my lead magnet I want it delivered instantly and the lead added to a nurture sequence, delivered as a precise step-by-step scenario for Zapier or Make. </context> <inputs> - Form or source: [TYPEFORM / WEBSITE FORM / ADS LEAD FORM] - The asset and where it lives: [PDF LINK / GOOGLE DRIVE / S3] - Email tool: [ESP NAME] - Nurture sequence to enroll in: [SEQUENCE / TAG] - CRM to log the lead in: [OPTIONAL] </inputs> <task> Build the scenario: trigger on form submit, validate the email, send the delivery email containing the download link, add the lead to the ESP with the nurture tag, log them in the CRM with the source, and alert sales if they match an ICP rule. Number each step and name the app plus action. </task> <constraints> - Real apps only; include the delivery-email subject and body. - Handle a bad or blank email (stop) and a repeat request (resend, do not duplicate). - Use a signed or expiring link if the asset is gated. </constraints> <format> Return the numbered scenario with the delivery email copy, then a note on how to A/B test the follow-up timing. </format>

You are a data engineer who writes production ETL jobs. <context> I need a scheduled job that ingests a CSV, cleans it, and upserts it into my database, delivered as one self-contained, commented Python script. </context> <inputs> - CSV source: [LOCAL PATH / URL / S3 KEY] - Target DB and table: [POSTGRES / MYSQL, TABLE NAME] - Column mapping and types: [CSV COL -> DB COL : TYPE] - Primary or unique key for upsert: [COLUMN(S)] - Cleaning rules: [TRIM, PARSE DATES, DROP BLANK ROWS, DEDUPE] </inputs> <task> Write a script that loads the CSV, applies the cleaning rules, validates types and required columns, connects to the DB, upserts rows on the key (insert or update), wraps the write in a transaction, logs counts (read, cleaned, inserted, updated, skipped), and rolls back on error. Include a requirements list and the crontab line. </task> <constraints> - One self-contained script; parameterized SQL (no string interpolation); read credentials from env. - Idempotent upsert; transactional; clear logging of every stage count. - Comment each stage. </constraints> <format> Return the full script in a code block, then the requirements and crontab line, plus a note on handling schema drift in the CSV. </format>

You are an automation engineer who builds Google Sheets integrations. <context> I want a scheduled sync that pulls data from an API into a Google Sheet, delivered as one self-contained, commented Google Apps Script I can paste into the Sheet's script editor. </context> <inputs> - API endpoint and auth: [URL, API KEY / BEARER / NONE] - Response shape: [PASTE A SAMPLE JSON OBJECT] - Fields to write and column order: [LIST] - Sheet and tab name: [NAMES] - Sync mode: [REPLACE ALL / APPEND NEW ONLY] - Schedule: [HOURLY / DAILY] </inputs> <task> Write an Apps Script that fetches the endpoint with the auth, parses the JSON, maps the chosen fields into rows in the column order, writes them to the tab in the chosen mode (clearing first for replace, or matching a key to append only new), and adds a time-driven trigger installer function for the schedule. Handle pagination if the API uses it. </task> <constraints> - One self-contained script; use UrlFetchApp and batch setValues (not row-by-row) for speed. - Store the API key in Script Properties, not inline; handle non-200 responses. - Append mode must not create duplicates. </constraints> <format> Return the full Apps Script in a code block, then a note on how to set the API key and install the trigger. </format>

You are a data-quality engineer. <context> My CRM has duplicate contacts and I need a script that finds and merges them safely, delivered as one self-contained, commented script. </context> <inputs> - Data source: [CRM API / EXPORTED CSV] - Match rules: [EXACT EMAIL / FUZZY NAME PLUS COMPANY / PHONE] - Merge rule: [KEEP OLDEST, FILL BLANK FIELDS FROM DUPES] - Fields to preserve: [LIST] - Output: [WRITE BACK VIA API / PRODUCE A MERGE CSV PLUS DELETE LIST] </inputs> <task> Write a script that loads the contacts, groups them by the match rules (normalizing case, whitespace, and phone format first), picks a surviving record per group by the merge rule, fills its blank fields from the duplicates, and outputs either the API updates or a merge-plan CSV plus a list of IDs to delete. Print a summary of groups found and records affected. </task> <constraints> - One self-contained script; NEVER hard-delete without producing the merge plan first (dry-run by default). - Normalize before matching; log every merge decision. - Deterministic survivor selection; comment each rule. </constraints> <format> Return the full script in a code block, then a note on how to review the merge plan before committing deletes. </format>

You are a DevOps engineer who automates backups. <context> I need a nightly job that backs up my database and uploads it to object storage with rotation, delivered as one self-contained, commented bash script. </context> <inputs> - Database: [POSTGRES / MYSQL, HOST, DB NAME] - Storage target: [S3 BUCKET / B2 / SPACES PLUS PATH] - Retention: [KEEP LAST N DAILY, N WEEKLY] - Compression: [GZIP / NONE] - Alert channel on failure: [SLACK WEBHOOK / EMAIL] </inputs> <task> Write a bash script that dumps the database, timestamps and compresses the file, uploads it to the storage path, verifies the upload succeeded, prunes backups older than the retention policy (local and remote), and sends an alert if any step fails. Include the crontab line to run nightly and a one-line restore command in comments. </task> <constraints> - One self-contained script; set -euo pipefail; read credentials from env or a protected file. - Verify the dump is non-empty before uploading; fail loudly with the alert on any error. - Never leave secrets in the process list; comment each step. </constraints> <format> Return the full script in a code block, then the crontab line and the restore command, plus a note on testing the restore. </format>

You are an analytics engineer who builds reporting pipelines. <context> I want a daily job that pulls metrics from several tools and writes a unified daily-metrics row to a warehouse table, delivered as one self-contained, commented Python script. </context> <inputs> - Sources and metrics: [E.G. STRIPE=MRR, GA=SESSIONS, CRM=NEW LEADS] - Warehouse and table: [POSTGRES / BIGQUERY, TABLE NAME] - Grain: [ONE ROW PER DAY] - Date to run for: [YESTERDAY / BACKFILL RANGE] - Failure behavior: [PARTIAL WRITE VS ALL-OR-NOTHING] </inputs> <task> Write a script that fetches each source's metric for the target date, assembles one unified row (date plus each metric), upserts it into the warehouse table on the date key so re-runs overwrite cleanly, supports a backfill loop over a date range, and logs which sources succeeded or failed. Handle a source being unavailable per the chosen failure behavior. </task> <constraints> - One self-contained script; parameterized upsert keyed on date; read credentials from env. - Idempotent (re-running a date overwrites, never duplicates); each source wrapped in try/catch. - Comment each source fetcher. </constraints> <format> Return the full script in a code block, then a note on how to add a new source and run a backfill. </format>