structural-retirement

Guides safe deprecation and deletion of APIs, features, services, tables, and code paths.

Updated May 27, 2026
One-click install
npx skills add https://github.com/ybaspinar/agent-work-skills --skill structural-retirement-ybaspinar
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: structural-retirement
Source: https://github.com/ybaspinar/agent-work-skills/tree/main/skills/structural-retirement
Command: npx skills add https://github.com/ybaspinar/agent-work-skills --skill structural-retirement-ybaspinar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Retiring software assets is risky: hidden consumers break, derived data keeps answering queries after the source is gone, and irreversible deletions happen without proof. This Skill structures retirement so removal only happens after live usage is proven zero and the irreversible step is separated from the decision. ## Core Features & Use Cases - Removal Proof: Requires live telemetry over the real usage cycle, dependency search, and identification of the slowest consumer before any deletion. - Reverse Contraction Sequence: Structures retirement as deprecate, switch reads, stop writes, observe cold, then drop, with every intermediate step runnable and reversible. - Decision/Act Separation: Splits the deletion decision (manifest, dry-run, review, four-eyes approval) from the irreversible delete act. - Use Case: When decommissioning a legacy API, use this Skill to produce a removal proof from telemetry, plan a Sunset/deprecation window, tombstone old consumers, and delete code, config, flags, docs, and dashboards together with a resurrection guard. ## Quick Start Use the structural-retirement skill to plan the safe deprecation and deletion of our legacy reporting API and its derived tables.

Frequently Asked Questions about structural-retirement

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

FAQPage Schema
How do I safely deprecate and remove an API?

Prove zero usage from live telemetry over the real usage cycle, then deprecate additively with a Sunset window for consumers you cannot force to upgrade. Contract in reverse: switch reads, stop writes, observe cold, then drop, keeping each step reversible.

How to prove a code path or table is unused before deleting it?

Use live usage telemetry observed over the full usage cycle, not grep, which is only a sample. Combine telemetry with dependency search and identification of the slowest consumer before treating anything as unused.

What is reverse contraction when retiring a service?

Reverse contraction means undoing the system in the opposite order it was built: switch reads away first, stop writes, observe the component cold, then drop it. Every intermediate step should be runnable and reversible.

Does soft-delete satisfy data deletion requirements?

No. Soft-delete is not destruction, and archiving cannot be used to avoid a duty to forget. Regulated or rights-bearing data requires honoring hard-delete obligations during retirement.

What are common mistakes when decommissioning features?

Common red flags include using a grep of the main repo as deletion proof, hard-cutting a public API with no deprecation window, fusing drop-and-rename with selection logic, and leaving dead docs, configs, or flags that still name the removed thing.