Browse Skills

MulmoTerminal Header configures session header buttons and chips for MulmoTerminal sessions. Use it to add shell actions, open links or overlays, and show live status like branch, diff counts, or context left.

MulmoTerminal Model configures MulmoTerminal to run on non-default models through providers, per-project provider/model settings, or custom launch commands. Use it when you need OpenRouter, a local or self-hosted model, or a different way to start Claude Code.

Recall Context shows which memory files are loaded and what memory files exist on disk in the current session. Use it to verify available context, missing files, and memory hooks before continuing work.

Recall Forget marks matching memory entries as superseded without deleting historical records. Use it when a stored fact is no longer current and needs to be kept with a supersession note.

Recall Maintain runs pressure-based cleanup for working memory when entries exceed budget. It also checks stale entries, contradictions, and open loops, and archives superseded items.

Recall Promote reviews recent daily logs and suggests entries to promote into CLAUDE.local.md or memory registers. Use it when turning recurring decisions, preferences, and project state into durable context.

Recall Search searches working memory, registers, daily logs, and archive files for relevant information. Use it when you need to find prior decisions, preferences, people, or past discussions.

Recall Status shows memory system health, including working memory size, stale entries, hooks, and file counts. Use it to check whether the recall memory structure is initialized and up to date.

Recall Write writes a note to the daily memory log after checking whether it changes future behavior, captures a commitment, or records a durable fact. Use it when you want only behavior-changing notes saved.

API Security Hardening helps secure REST and GraphQL APIs against common attack vectors. Use it when adding authentication, validating input, rate limiting, handling uploads, or exposing endpoints to outside consumers.

Deep Research produces a cited Markdown report from broad parallel web research, multi-source validation, and confidence tracking. Use it for thorough analyses, comparisons, landscapes, literature reviews, and other questions that need evidence from many sources.

Deep Research structures a Claude Code research run around primary sources, comparison, validation, and a single final report. Use it when you need a defensible judgment instead of a source roundup.