enough

Two sibling visualization jobs in one skill. GIRRAPH — IBIS discipline for mapping contested, fuzzy, or wicked problems as issues (❓), positions (💡), and arguments (➕/➖) in a .girraph file, BEFORE any solving happens; engage when the user wants to map a problem, untangle a decision with no clean answer, see the shape of a disagreement, or says 'girraph this', 'map this out', 'help me think through (not solve) X', 'what are the positions on Y', 'IBIS', 'argument map'. MERIRMAID — generate, display, and edit Mermaid diagrams in .merirmaid files for ANYTHING with structure or flow; engage when the user says 'diagram this', 'flowchart', 'mermaid', 'sequence diagram', 'state diagram', 'visualize this process', 'draw the architecture', 'show how X works', 'chart the steps'. Also engage when a planning conversation keeps circling — competing considerations that won't linearize are a girraph smell; a described-but-unpictured process or structure is a merirmaid smell. Do NOT engage for problems with a known procedure

Three-mode analytical toolkit for documents, books, web pages, and decisions. SUMMARIZE produces an even-handed one-page digest of any text (author motivation, intended audience, tone, key quotes). PROOFREAD makes light spelling/typo corrections on full-length documents (including entire books, chapter by chapter), driven by the Harper grammar checker (local, offline, rule-named findings), and produces a separate proof report with suggestions and repeated-phrase findings. DECIDE picks three archetypal personas from a built-in roster and runs a transcripted debate over a question, complex decision, or 'what should I do next' — returning a recommendation plus the full debate. Use for 'analyze this', 'summarize this', 'one-pager', 'tl;dr but smart', 'read this for me', 'proofread this', 'copy-edit this book', 'fix typos in', 'spot repetition', 'help me decide', 'pros and cons', 'what should I do about', 'three perspectives on', 'debate this'. Works with .txt, .md, .docx, .pdf, .epub, .html, pasted text, web page

Multi-session memoir planning and (optionally) drafting via patient, iterative dialogue. Use whenever the user wants to plan, build, or write a memoir — full life, partial life, single milestone, professional thread, creative thread, or thematic slice. The agent interviews the user across many sessions, accumulates plan documents in a project folder, and optionally synthesizes a draft. Trigger on memoir-dialectic, "help me plan my memoir," "write my memoir with me," "memoir interview," "life story project," "autobiography help," or any long-horizon autobiographical writing collaboration.

Pre-installation explainer and security/epistemic auditor for skill packages (.skill zips or skill directories). For any skill the user is considering installing, produces a single report that LEADS with a plain-English 2–3 paragraph description of what the skill does and why someone might want it, then follows with the safety verdict — prompt-injection vectors, privilege escalation patterns, epistemic pathologies, and executable payload risk. Use whenever a user wants to understand, vet, audit, inspect, or safety-check a skill before installing it, when downloading skills from untrusted sources, when onboarding a skill into an existing ecosystem, or on any request like "what does this skill do," "explain this skill," "scan this skill," "is this skill safe," "audit this zip," "vet this skill," "check this skill for injection," "skill-scanner," "skill security check," "is this skill malicious," or "should I install this." Also trigger when a user mentions downloading skills from the internet, skill marketplace

Offline translation across ~419 languages via MADLAD-400-3B-MT (CTranslate2 + SentencePiece). No cloud. No account. The model downloads once (~3 GB) and runs locally on CPU or Apple Silicon thereafter. Use whenever the user wants to translate text TO or FROM a non-English language — short phrases, paragraphs, whole documents, idioms, low-resource languages, indigenous languages, dead languages MADLAD has seen, polite forms, casual registers. Also trigger on "translate", "say this in <lang>", "what does <foreign text> mean", "render this in <language>", "<lang> for <phrase>", "how do you say X in Y", "localize", or any request that involves moving text between human languages. DO NOT use for code translation between programming languages. DO NOT use for transliteration (script conversion) without a target meaning — MADLAD translates meaning, not glyphs.