From memory-os
Use when starting work in a project to detect context and propose the right optional tool — CodeGraph, Understand-Anything, taste-skill, Firecrawl, or a knowledge layer. Proposes; the user confirms.
How this skill is triggered — by the user, by Claude, or both
Slash command
/memory-os:situational-suggestionsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Detect the situation, then PROPOSE the matching tool and let the user decide. Never enable or install silently. Ask once per situation; if declined, do not nag.
Detect the situation, then PROPOSE the matching tool and let the user decide. Never enable or install silently. Ask once per situation; if declined, do not nag.
| Situation detected | Propose | Note |
|---|---|---|
Code work in a repo with .codegraph/codegraph.db present | Use codegraph_* MCP queries instead of file sweeps | Automatic for code — no prompt needed; it only saves tokens. |
Code work, but no .codegraph/ present | "Want me to set up CodeGraph for token-efficient code queries?" | Agent-facing. |
| You land in an existing/unfamiliar project | "Want me to run Understand-Anything to build a comprehension graph + onboarding tour?" | User-facing comprehension (complements CodeGraph). |
| Task involves designing or redesigning a site/UI | "Want to apply taste-skill design guidance? Set the dials: Design Variance / Motion Intensity / Visual Density, and a direction (minimalist, brutalist, soft…)." | Avoids generic AI design. |
| Starting a new project from an idea, or a URL worth ingesting | "Want me to research/ingest with Firecrawl (URL → clean markdown → .memory knowledge)?" | Requires a Firecrawl key. |
Domain knowledge is piling into .memory/knowledge/raw/ | "This is accumulating knowledge — want me to stand up a typed-edge wiki layer (raw/wiki/outputs + index)?" | Never auto-created. |
Project has .memory/ but no IDEA.md, and acronizer.config.json is present on this machine (check $ACRONIZER_CONFIG env var or run find ~ -name acronizer.config.json -maxdepth 10 2>/dev/null) | "Want to add this project to your Acronizer pipeline? Run /acronizer-add." | Agent-facing. Propose once per session; if declined, do not repeat. |
Project has .memory/ and STATE shows a phase or the Definition of Done met (a runnable baseline exists) and an objective dimension is in play (tests pass but could be faster, bundle size, latency, cost, a benchmark) | "You've got a working baseline and something measurable — want to spin up an Auto Research loop (/auto-research) to optimize it overnight?" | Agent-facing. Propose once per situation; if declined, do not repeat. |
All of these are optional and off by default (except CodeGraph detection, which is automatic for code because it has no downside).
npx claudepluginhub tpop78/memory-os --plugin memory-osProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.