- Home
- Skills
- Testing & Debugging
- Automation Delivery Test Pack
Automation Delivery Test Pack
Design acceptance and recovery tests for automations to handle duplicate triggers, partial failures, and retries.
$5
Works with the AI Agents you already use
Automation Delivery Test Pack
Example session with this skill installed
I have a workflow: Webhook Trigger -> Create Customer in CRM -> Send Welcome Email. I need a test pack for this. CRM creation is not idempotent. Email tool handles retries automatically.
- Read your context and instructions
- Compiled the automation delivery test
I have designed the test pack for the Webhook-CRM-Email workflow. Because CRM creation is not idempotent, the recovery plan focuses on manual reconciliation and lookup gates to prevent duplicate records.
Test Matrix and Recovery Plan
| Test ID | Scenario | Stimulus | Expected System State | Evidence Required |
|---|---|---|---|---|
| TC-001 | Normal Path | Unique Webhook Event | 1 CRM Record, 1 Email Sent | CRM Entry Timestamp, Email Log |
| TC-002 | Duplicate Trigger | Replayed Webhook ID | 1 CRM Record (No Change), 0 New Emails | Deduplication Log / CRM ID Check |
| TC-003 | CRM Partial Failure | Webhook (CRM 500 Error) | 0 CRM Records, 0 Emails | CRM Error Log, Checkpoint Status |
| TC-004 | Post-Write Timeout | Webhook (CRM Success, Timeout) | 1 CRM Record, 1 Email | CRM Search by Event ID, Log Check |
Recovery and Reconciliation Logic
- CRM Write Failure: If a 500 error or timeout occurs, do NOT auto-retry.
- Lookup Gate: Before any replay, perform a lookup in the CRM using the
event_idoremailas a key. - Branching: If record exists, skip to Email step. If record does not exist, proceed with CRM creation.
Email Idempotency: Rely on the email provider's built-in retry logic as specified; however, log the message_id for reconciliation.
Required Acceptance Evidence
- Deduplication Check: Prove that a second webhook with the same ID does not create a second CRM record.
Atomic Recovery: Demonstrate that a CRM timeout followed by a manual reconciliation lookup correctly identifies the created record before triggering the email.
Next steps
- Provide the specific field names used for the CRM lookup to finalize the synthetic inputs.
- Define the
event_idretention window for the deduplication scope. - Review the Optional Test Follow-up for concurrent webhook delivery scenarios.
Connects securely to your tools. The creator never sees your data.
About this skill
Deploying custom automations without a rigorous test plan leads to duplicate triggers, partial failures, and data corruption. This skill generates a comprehensive acceptance and recovery test pack based on your workflow maps. It moves beyond simple "happy path" testing by identifying failure branches, deduplication gaps, and recovery prerequisites.
What it does
- Test Design creates a matrix covering normal paths, duplicate events, and partial failures after writes.
- Evidence Review reconciles run evidence against acceptance criteria to identify unverified steps.
- Synthetic Input Generation produces stable test IDs and controlled stimuli for safe environment testing.
- Recovery Logic Mapping defines retry guidance, idempotency checks, and manual reconciliation paths for unknown states.
- Client Acceptance Sheets generates plain-language reports distinguishing between requirements, optional tests, and unresolved policy decisions.
How it works
- Intake the workflow map, expected business outcomes, and any available run evidence.
- Map each step to its success signals, failure branches, checkpoints, and deduplication scope.
- Generate a test matrix with specific prerequisites, stimulus, and expected per-system state counts.
- Evaluate evidence to mark criteria as Demonstrated, Contradicted, or Unverified.
- Report final status as Hold for defect, Pending evidence, or Ready for client review.
Frameworks & tools
Works with any workflow orchestration tool (Zapier, Make, n8n, AWS Step Functions) by analyzing the logical workflow map and system effects.
Why this beats prompting it yourself
Generic prompts often overlook the "partial failure" state where a write occurs but the response times out. This skill enforces rigorous idempotency and reconciliation logic that most developers forget to document, ensuring your automations don't trigger double payments or duplicate records.
Use cases
- Validating a new Shopify-to-ERP sync before moving to production.
- Reviewing logs from a failed automation to determine if a retry is safe.
- Creating a standardized UAT pack for a client-facing delivery.
- Analyzing high-concurrency workflows for potential race conditions and rate limits.
Known limitations
Does not perform production writes or API calls. Requires a workflow map or run evidence to be provided in text/markdown format.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Does not execute production actions or perform live API calls. Requires manual provision of workflow maps or logs. Scope is limited to logical verification and recovery planning.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 7 days ago