harden

Hardens error handling, network failure recovery, and edge case resilience across a codebase.

4|Updated Mar 14, 2026
One-click install
npx skills add https://github.com/VoxTechnologies/transmute-framework --skill harden-voxtechnologies
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harden
Source: https://github.com/VoxTechnologies/transmute-framework/tree/main/skills/harden
Command: npx skills add https://github.com/VoxTechnologies/transmute-framework --skill harden-voxtechnologies

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Products built feature-by-feature often have cross-cutting failure modes that individual implementations miss: network drops mid-operation, race conditions from concurrent edits, silent error swallowing, and retries that duplicate data. This Skill systematically audits and hardens the entire product for these resilience gaps. ## Core Features & Use Cases - Multi-Agent Resilience Audit: Coordinates three specialized teammates covering backend resilience, frontend error UX, and cross-cutting edge cases. - Retry, Timeout, and Circuit Breaker Patterns: Applies exponential backoff with jitter, calibrated timeouts, idempotency verification, and circuit breaker decision logic for external service calls. - Edge Case and Concurrency Hardening: Handles concurrent edits, empty/boundary states, session expiry, deep links, and optimistic update rollbacks, with E2E resilience tests. - Use Case: After implementing all product features, run this Skill to verify every external API call has timeout and retry handling, every form preserves user input on failure, and multi-step operations cannot leave inconsistent data. ## Quick Start Ask the agent to harden error handling and add resilience across the product, including retry logic, circuit breakers, and edge case coverage.

Frequently Asked Questions about harden

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add retry logic to API calls without duplicating data?

Verify idempotency before adding retries: only idempotent mutations get automatic retries with exponential backoff. For non-idempotent operations, add an idempotency key based on a natural deduplication key, or surface errors to the user with a manual retry option.

How do I harden error handling across a full-stack application?

Split the work across three scopes: backend external service failures and data consistency, frontend error boundaries and network failure UX, and cross-cutting edge cases like concurrency and boundary conditions. Run the full test suite afterward and produce a gate-decision report.

When should I use a circuit breaker versus retry with backoff?

In serverless environments, retry with exponential backoff alone is the recommended default since there is no persistent in-memory state. Use full circuit breakers only for long-running servers or external services with frequent prolonged outages, and always include a reset mechanism.

What timeout values should I use for external API calls?

Use 3-5 seconds for fast API calls, 30 seconds for file operations, and 60-120 seconds for AI model calls. If a performance report with measured latencies exists, use the measured value plus a 50 percent buffer instead of generic baselines.

Why does retrying mutations cause duplicate records?

Mutations that increment counters, append logs, or send notifications are not idempotent even when they use set operations, so side effects accumulate on each retry. Classify mutations first and only add automatic retries to truly idempotent create or update operations.