wellmate-memory

Records durable architecture decisions and project facts to prevent rediscovery in future work.

Updated Jan 1, 2026
One-click install
npx skills add https://github.com/sarkarshivaditya-lab/WellMate --skill wellmate-memory-sarkarshivaditya-lab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: wellmate-memory
Source: https://github.com/sarkarshivaditya-lab/WellMate/tree/main/.engineering-skills/wellmate/memory
Command: npx skills add https://github.com/sarkarshivaditya-lab/WellMate --skill wellmate-memory-sarkarshivaditya-lab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Engineering context is lost between sessions, forcing repeated rediscovery of architecture decisions, locked systems, and validation constraints. This Skill maintains a durable project memory file so future work starts from established facts instead of re-investigating the codebase. ## Core Features & Use Cases - Durable Fact Capture: Records architecture decisions, canonical data ownership, fragile subsystems, rollback points, and validation requirements in PROJECT_STATE.md. - Change Discipline: Requires reading existing memory and comparing it against current code before significant changes, updating only when durable facts change. - Failure Documentation: Captures root causes, affected subsystems, resolutions, and validation performed after important failures. - Use Case: Before modifying the emergency detection module, read the recorded state model and validation limits, then update memory only if the deterministic behavior materially changes. ## Quick Start Read the project memory file and record any durable architecture or validation facts from the current work session.

Frequently Asked Questions about wellmate-memory

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

FAQPage Schema
How do I maintain project memory across AI coding sessions?▼

Record durable facts like architecture decisions, locked systems, and rollback points in a dedicated memory file such as PROJECT_STATE.md. Before significant changes, read existing memory, compare it against current code, and update only when a durable fact changes.

What should be recorded in an engineering project memory file?▼

Record architecture decisions, canonical data ownership, fragile subsystems, known regressions, dependency behavior, validation requirements, rollback points, and rejected approaches. Exclude temporary thoughts, trivial details, secrets, credentials, and facts trivially rediscoverable from code.

Does project memory override the source code or documentation?▼

No. Memory never overrides source code or primary docs like CLAUDE.md. When memory conflicts with the repository, inspect the repository, resolve the discrepancy, and then update the memory to match verified reality.

When should I update project memory after a bug fix?▼

Update memory after discovering an important failure by recording the root cause, the affected subsystem, the successful resolution, and the validation performed. Skip fixes that are trivial or do not change durable architecture facts.

What are the limitations of a file-based project memory approach?▼

Memory can drift from reality if not verified against the current code, and it cannot capture runtime validation that never occurred. It requires discipline to read before changes and honesty about unverified capabilities such as browser or native testing.