- Home
- Skills
- APIs & Backend
- Webhook Retry & Recovery Review
Works with the AI Agents you already use
Webhook Retry & Recovery Review
Review duplicate actions and lost work in webhook handlers. Build reproducible retry, concurrency and recovery checks with a seven-scenario local Node.js harness.
$5
Webhook Retry & Recovery Review
Example session with this skill installed
Synthetic example: a receiver permanently marks an event accepted before creating its required archive record. Review it without changing the source; reproduce a failure before archive commit using a fake destination, then check normal and legitimate-event controls.
- Read your context and instructions
- Compiled the webhook retry
Illustrative finding: the early accepted mark prevents recovery after a pre-commit exception; redelivery returns without creating the required record. Include source location, fault schedule, missing-effect evidence, a scoped fix proposal and a regression asserting eventual completion. This is a synthetic illustration, not a customer result.
Connects securely to your tools. The creator never sees your data.
About this skill
A duplicate check can stop the second email and still lose the first one. A successful response can hide an action that never completed.
Webhook Retry & Recovery Review guides your agent through the boundaries that matter: event identity, atomic claims, the external action, completion records and recovery. It helps turn a suspicious code path into a reproducible finding and a focused regression test.
- Review both repeated actions and work silently suppressed by deduplication.
- Check legitimate events on the same object and independent tenant scopes.
- Document the precise failure window, observed effect and proposed fix.
- Use an optional dependency-free Node.js replay harness for seven local event-action scenarios with a fake destination.
Designed for developers reviewing an existing handler. The harness requires an inspected, trusted adapter and disposable test environment; it executes local code with normal permissions. It does not connect to payment accounts or repair live workflows. Its synthetic destination and injected exceptions do not certify real provider, database or n8n behavior. No exactly-once, security or time-saving guarantee.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for APIs & Backend.
Fixed per-event-action helper, not a universal state-reconciliation oracle. Trusted adapter execution is not sandboxed by the helper. Fake sink idempotency must match the real destination; exceptions are not process/database crashes. No production-provider, signature, full-app security or independent customer benefit certification. Marketplace delivery remains untested.
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 15 days ago
- Software
- Trusted JavaScript adapterDisposable test environmentPer-event-action business contractProject-specific tests
- Optional
- Terminal access
Node.js 22+
Node.js runs the optional replay helper without npm dependencies; the contract must match it. The test environment uses fake external actions. Tests cover state reconciliation.
Permissions
File Scopes
No provider accounts, production events, email sending, payment execution, package installs or deployment are required. The helper itself does not use the network, but it imports trusted adapter code; that code and its dependencies must be inspected and disconnected from real services. Agent/model-provider data handling is separate from keeping the test effects local.