skills

Use this skill when writing nodes into an NKS graph — creating phenomena, kriyas, vimarshas, holons, or kartas. Triggers: 'запиши', 'зафиксируй', 'добавь узел', 'создай крию', 'create phenomenon', 'create entity', 'add to graph', 'nks_add', 'положи в граф', or whenever the agent is about to call any nks_add_* tool or nks_batch with creates. Also when unsure which node type, given_as, or modes to pick. Use even for single-node writes — one wrong type propagates through the graph. Distinct from design (plans paths from goals), weaving (repairs existing structure), and methodology-work (evolves methodology itself).

Use this skill to build a product roadmap for a GitHub repository — or a multi-repo product (a whole org, or a set of repos forming one product) — derived from its issues and PRs through an NKS reasoning graph. Triggers: 'roadmap from repo', 'build a roadmap from issues', 'roadmap from github', 'roadmap for my org', 'multi-repo roadmap', 'roadmap across repos', 'quick roadmap', 'roadmap teaser', 'собери роадмап из issues/PR', 'роадмап по продукту', 'что в этом репо делать дальше', product-roadmap. Role-plays the maintainer: verified present-state ground, weighted backlog as shabda, directions underway (bianhua), rendered as graph + markdown + local HTML. Composes verstakify, intake, and assembly. Needs the gh CLI and nks_* MCP tools.

Use when the user asks to verstakify a repo — bootstrap or refresh its AGENTS.md / CLAUDE.md to the verstak standard, apply the NKS methodology conventions, or wire the session-lifecycle rituals (orient-in-NKS on start, push→update-NKS hooks, quality gate, memory guard, CLAUDE.md pointer — @AGENTS.md import, Windows-safe). Triggers: "verstakify", "verstakify this repo", "привести проект к стандарту", "завести/обновить AGENTS.md", "set up AGENTS.md", "bootstrap AGENTS", "apply the meta template", "наведи порядок в конфиге агента". AGENTS.md is a derived view, not hand-written prose: each concern is audited against its source of truth and re-projected when stale, preserving authored judgment; fresh repos ask the user for the authored slots. Needs the nks_* MCP tools for the NKS steps.

Use this skill to stand watch — the agent's duty cycle in a verstakified repo, autonomous work from the doer's inbox agenda. Triggers: 'заступай на вахту', 'вахта', 'разгребай инбокс', 'on duty', 'stand watch', 'sweep the inbox', 'duty cycle', or an autonomous session start where AGENTS.md names an agent karta. Also scopes to one transformation: 'продвигай bianhua', 'доведи превращение', 'drive the bianhua' — the queue becomes its open anga, arrival is proposed to the owner. Loop: orient → agenda → pick workable vimarshas → work per repo ritual (commit, push, PR) → wire dependencies into other doers' inboxes → weave the wake → wait (webhook or ≤30-min re-check) → escalate transcendent will to the owner. Composes entry, inquiry, writing, weaving; the graph is the durable state between wakes. Needs nks_* MCP tools and a verstakified AGENTS.md.

Use this skill BEFORE responding to any question that touches: methodology, knowledge graphs, NKS, conceptual frameworks built around vimarsha (open inquiry), kriya (process/practice), holon (boundary), karta (role with motivation), phenomenon (что-то, на что действуют крии — formerly "entity"), estafeta (relay/handoff). Use it when the user references "the graph", "in NKS", "in methodology", "by my notes", "we discussed", "what does the realm say", or asks "what do I think about X" in a context where they have prior recorded thinking. Use it when the user references named concepts that sound like established terminology in their own work — even if you don't recognize the term, the realm probably does. When uncertain — use it; orient is cheap, answering from training data about a topic the realm has worked through is expensive and often wrong.

Use this skill for the LIFE of a vimarsha and work with the FIELD of inquiry — not posing one (that's writing), but moving it through its lifecycle: anchoring, resolving, closing, parking, superseding, crystallizing, or attaching it to a bianhua. Triggers: 'разбери вимарши', 'поле вопрошания', 'закрой вимаршу', 'запаркуй вимаршу', 'разреши вопрошание', 'триаж вимарш', 'work the field of inquiry', 'close/park this vimarsha', 'resolve the question', a high dangling-vimarsha count in orient, or session close. Distinct from writing (which poses a vimarsha — genre, name-question) and from assembly (the realm-wide field pass): this skill is the per-vimarsha and field work of carrying questions to a resolution. Needs the nks_* MCP tools.

Use when the user asks to design, architect, or project a system, component, or feature and NKS tools are available. Triggers: 'спроектируй', 'давай продумаем', 'архитектура', 'нужен дизайн', 'design', 'let's think through', 'plan the system', 'what should this look like', mentions of 'жц' / 'lifecycle', 'крия' / 'эстафета' / 'холон', or 'записать в NKS' / 'положить в граф' / 'зафиксировать'. Also use at decision moments — 'we chose X', 'we decided', 'persist this design' — and right after a brainstorming/elicitation session produced a design or spec: the graph is where it becomes the record. Also use when the conversation turns from reading to building ('what should we build?'). Use even for seemingly simple design tasks — the graph prevents losing decisions, questions, and risks.

Use this skill to assemble a realm — the central ritual of the 時-cycle: discern the pattern over a whole field of inquiry and activity and produce 形, the assembly map. Triggers: 'собери реалм', 'сборка', 'пересборка', 'что здесь происходит по сути', 'повестка', 'состояние реалма целиком', 'assemble the realm', 'what's really going on here', 'reassemble', 'map the transformations'. Also use when entry/orient shows a large ungrouped field — many free vimarshas, a wall of top-level kriyas. Distinct from weaving (semantic repair of existing structure), design (building paths from goals), and inquiry (the per-vimarsha lifecycle work). Needs the nks_* MCP tools, especially nks_add_bianhua and lens="bianhua".

Use this skill to bring EXTERNAL WORD into an NKS realm — the agent-facing shabda-intake. Triggers: 'впусти внешнее слово', 'засей граф из issue/доков', 'внеси issue в граф', 'intake', 'ingest issues into the graph', 'seed the realm from docs/README/conversation', 'bring external word into NKS', 'shabda intake'. Source-independent: issues, docs, code, a conversation enter through an adapter under one discipline — form→node type, epistemic mode by kind of source, provenance source_kind=shabda, dedup, anchor to the source, verify and graduate. Distinct from writing (one node from a distinction you already hold), verstakify (projects derived config), and product-roadmap (composes this layer with a GitHub adapter). Needs the nks_* MCP tools.

Use this skill for a transformation's integrity — right after a bianhua is created (telos accepted by the owner) or on request: 'integrity check', 'impact analysis', 'what will this change touch', 'проверь бяньхуа на целостность', 'что затронет это превращение', 'кого заденет' — and for the reverse claim-audit when a prose artifact claims graph-backed leverage: 'is this backed by the graph', 'claim audit', 'подкреплено ли графом', 'не театр ли'. Forward: propagate the telos, pose clustered «is this affected?» samshayas on the wavefront. Reverse: verify each structural claim is carried by real edges, report claimed-but-unwired. Distinct from assembly (map from the field) and design (paths from goals). Needs the nks_* MCP tools.

Use this skill when doing semantic work on an existing NKS graph: closing lifecycles, writing sense on arrows, splitting kriyas with multiple actors, reconnecting arrows after phenomenon distinctions, or any work where the graph exists structurally but needs semantic completeness. Triggers: 'проткать', 'прошить', 'ткачество', 'weave', 'close lifecycle', 'fix tensions', 'sense на стрелках', references to leaked/relay-gap/orphan tensions, or when nks_orient(lens="tensions") shows structural problems. Also triggers when entry shows a realm with high tension count (large leaked/relay-gap sections in lens="tensions"). Distinct from design: design creates structure from goals, weaving completes structure that already exists.

Use this skill when working on the methodology realm itself — not applying methodology to a project, but evolving the methodology. Triggers: 'methodology realm', 'работа над методологией', 'обогатить методологию', 'methodology нужно дополнить', 'записать принцип', 'новая крия в methodology', 'новая нить', references to 'methodology' realm by name, or when the user brings agent feedback about methodology gaps. Distinct from design (designing external systems) and weaving (semantic repair). This is meta-work: methodology applied to itself. The stakes are highest — every node here shapes how every future agent thinks.