jpj-memory-manager_skill

Overview

This skill automatically captures and organizes session-end changes to infrastructure, projects, and tools. It produces concise daily logs and updates long-term memory files only for structural or credential changes. The goal is consistent documentation without bloating memory with transient details. Use it to keep operational history accurate and searchable.

How this skill works

At the end of a session the skill scans the conversation for keywords related to servers, deployments, cron jobs, repositories, APIs, integrations, and credentials. It filters out noise such as transient tests, debugging steps, and minor tweaks, then writes a timestamped daily log summarizing meaningful changes. If a change is structural or long-term (new project, architecture decision, new credentials, server), it updates MEMORY.md, TOOLS.md, or PROJECTS.md according to the defined criteria. The output format is short, factual, and designed for quick review and future reference.

When to use it

• End of any session that changed servers, services, deployments, or cron jobs
• After project lifecycle events: new project, status change, or major feature deployment
• When adding or rotating credentials, API keys, or new integrations
• After infrastructure fixes, migrations, or configuration changes
• When design or architecture decisions were made that must persist across sessions

Best practices

• Run only at session end to avoid partial or noisy logs
• Keep daily log entries concise: 1–2 sentence summary and short bullet list
• Only add to MEMORY.md for long-term structural changes; keep those entries under 5 lines
• Record credentials and new servers in TOOLS.md with essential metadata only (purpose, IP, access method)
• Avoid saving debugging steps or temporary tests unless they produced an important lesson

Example use cases

• Deployment completed: record service deployed to specific server and note any migration steps in the daily log
• Cron update: summarize changed schedule and script purpose, log impact if any
• New project kickoff: add project entry to PROJECTS.md and note initial stack and deployment URL
• Credential onboarding: save new API key entry in TOOLS.md with service name and usage notes
• Architecture decision: write a short MEMORY.md entry explaining the decision and rationale for future reference