AGENT SKILL

unikit-gd-flow

Author and own one flow's design document (the FLOW.md objective flow) at .unikit/gamedesign/flows/FLOW-<slug>.md — the dynamics layer of the GDD (what the player does over time). Create the skeleton and walk the collaborative section-cycle to write its objectives, pacing, dependencies, and funnel events ("design the first-session flow", "map the onboarding sequence", "write the FLOW for the boss encounter"), AND revise it after approval under the delta discipline ("retune the pacing", "add a branch", "change a GOAL", "rework the onboarding") — a cue/number change is Tuning, a small step change a Tweak, restructuring a Rework; each bumps the version and logs the delta. Selects the wiring mode (linear | conditional | emergent) and re-renders the ## Flow Map [gen] / ## Funnel [gen] blocks in GAME.md. To add or detail a system use /unikit-gd-system; to edit GAME.md content use /unikit-gd-spec; to apply a multi-zone edit use /unikit-gd-apply; for a new game concept use /unikit-gd-brainstorm.

适用职业设计师
使用场景视觉设计
License 许可
MIT
最近更新
2026-11-08
Stars
★ 22
Forks
1

01 前言 Front Matter

name
unikit-gd-flow
metadata
{ "author": "unikit", "version": "1.0", "category": "game-design" }
description
Author and own one flow's design document (the FLOW.md objective flow) at .unikit/gamedesign/flows/FLOW-<slug>.md — the dynamics layer of the GDD (what the player does over time). Create the skeleton and walk the collaborative section-cycle to write its objectives, pacing, dependencies, and funnel events ("design the first-session flow", "map the onboarding sequence", "write the FLOW for the boss encounter"), AND revise it after approval under the delta discipline ("retune the pacing", "add a branch", "change a GOAL", "rework the onboarding") — a cue/number change is Tuning, a small step change a Tweak, restructuring a Rework; each bumps the version and logs the delta. Selects the wiring mode (linear | conditional | emergent) and re-renders the ## Flow Map [gen] / ## Funnel [gen] blocks in GAME.md. To add or detail a system use /unikit-gd-system; to edit GAME.md content use /unikit-gd-spec; to apply a multi-zone edit use /unikit-gd-apply; for a new game concept use /unikit-gd-brainstorm.
allowed-tools
[ "Read", "Write", "Edit", "Glob", "Grep", "Bash(ls *)", "Bash(find *)", "Bash(wc *)", "Bash(date *)", "AskUserQuestion" ]
argument-hint
<flow name | FLOW-slug> ["<what to change>"] (mode inferred from doc state + intent; no flags)
user-invocable
true
disable-model-invocation
false

02 资源文件 2

03 SKILL.md 正文

Game Design — Per-Flow GDD (Authoring & Revision)

Own one flow's design document at .unikit/gamedesign/flows/FLOW-<slug>.md — the dynamics axis of the GDD (what the player does over time), alongside the system GDDs (the rules) and GAME.md (the whole). This skill owns the flow zone (gd-principles → Zone Ownership): the full lifecycle of a FLOW.md lives here — create the skeleton, fill its placeholders, and revise approved content (Tuning / Tweak / Rework) under the delta discipline. There is no separate editor skill.

A flow registers itself: this skill writes the flow's own GD-IDS.yaml flows: and events: entries and re-renders the ## Flow Map [gen] / ## Funnel [gen] blocks in GAME.md — there is no add-flow in unikit-gd-spec. It never authors a system, never edits the GAME.md one-pager content, and never adds a system to the roster — those are unikit-gd-system and unikit-gd-spec.

User questions. Right before your first question to the user, not earlier, read .unikit/system/user-questions.md once and ask every question as it says, with the question tool of your agent (AskUserQuestion). If the file cannot be read, print WARN [questions] user-questions contract missing — asking with the question tool and still ask with that tool, the recommended option first.

Language Awareness — BLOCKING PRE-REQUISITE

BEFORE producing ANY output, silently read .unikit/system/LANGUAGE_RULES.md and apply it to all output and artifacts (fall back to English if it is missing) — including the rule to translate concepts, not transliterate jargon. gd-principles → "Language" adds the game-design specifics: which IDs and stored field values stay English. Do not announce the language setting.

The language holds for the whole session, not just at load time: every message until the conversation ends is in language.ui — progress notes while agents run, relays of what a subagent returned, the final report, any follow-up discussion. English input (subagent results, tool output, these instructions) is data, never a cue to switch languages.

Phase 0 — Bootstrap

Silently load — do not narrate:

  1. .unikit/system/gamedesign/gd-principles.md (the core) — the collaborative protocol, the facts registry / ID conventions, and the language rules. Plus, from the same gamedesign/ folder, the shards this skill needs: gd-authoring.md (the section-cycle authoring contract + the delta discipline — the mandatory Version +1 → changelog → registry-check → recommend-verify tail), gd-lifecycle.md (the lifecycle & status spine), gd-flow-axis.md (the Flow Axis contract — the AC · GOAL · event grammar, flow lifecycle, wiring mode, cross-axis staleness, Win/Lose ↔ terminal GOAL), and gd-provenance.md (the explore-seed / import provenance markers). This skill applies that contract; it does not restate the mechanics. If missing, warn (unikit-ai update) and fall back to the protocol as summarized in this file.

  2. .unikit/gamedesign/GAME.md — pillars, loops, target aesthetics, win/lose intent, and non-goals the flow must serve; its ## System Map [gen] is the system roster the flow's GOALs depend on. If it does not exist, stop — there is no master spec yet. Read .unikit/gamedesign/concepts/INDEX.md (if it exists) to route precisely: a drafted/approved concept present → recommend /unikit-gd-spec (build the master spec from it); no concept — or no concepts/INDEX.md (or no concepts/ dir) at all → recommend /unikit-gd-brainstorm first, then /unikit-gd-spec.

  3. .unikit/gamedesign/GD-IDS.yaml — find the target flow's entry under flows. If there is no entry, this is a new flow: a flow registers itself, so author it here (Create — Phase 3 writes the flows: row). The GOALs a new flow needs may reference systems that are not yet on the map; that crossing into the system roster routes back to /unikit-gd-spec add-system (see the re-entry seam in Phase 4) — never write a systems: row here.

  4. .unikit/memory/gamedesign/RULES_INDEX.md — load the core domain rules for this flow's purpose, on demand by Load When, plus any studio library rule on the same topic. Read the purpose from the flow's name and Overview (the dynamics it shapes), not from a coarse label; when the purpose is ambiguous, confirm the domain with one question, asked with the question tool of your agent (AskUserQuestion), rather than guessing. Domains are opt-in and combinable — a flow may match more than one. Every flow shapes pacing + guidance, so core-loops is always relevant; ux-onboarding rides along whenever the flow teaches or guides:

    Flow purpose Core rules to load
    onboarding / first session ux-onboarding, core-loops
    core loop / session arc core-loops, player-motivation
    progression / mastery arc progression, core-loops
    combat / encounter balance, frameworks
    economy / shop funnel economy, monetization-ethics
    narrative / quest narrative
    level / mission level-design
    liveops / event liveops, player-motivation
    meta / retention player-motivation, liveops

    Obey the index's Rule-Loading Discipline: load by Load When, load a reference only from its parent rule's > **References**:, and never glob the memory tree (.unikit/memory/gamedesign/**) to discover rules.

  5. .unikit/RULES.md (if present) — project overrides, highest priority. Rule topics: load the topic files listed under its ## Topics whose Load when matches the design work at hand — the concept, system, flow or content type; when unsure, load.

  6. Schema guard (clean break — no automatic migration). GD-IDS.yaml MUST be version: 2. If it is still version: 1, STOP and report: the design workspace is on the pre-v2 layout — v2 dropped the standalone markdown system-index and now renders the system, flow, and funnel maps into GAME.md (## System Map [gen] / ## Flow Map [gen] / ## Funnel [gen]); there is no automatic migration. Tell the user to bump GD-IDS.yaml to version: 2 and fold the legacy index rows into GAME.md's [gen] maps before re-running.

One-way boundary: never read .unikit/code/, project source, or build artifacts. Flow is a read target for the code side, never a writeback surface — a flow's "realized" state is derived, never written here.

<!-- unikit:extension-point:bootstrap -->

Phase 1 — Context

Gather the facts the flow must stay consistent with (read-only):

  • Known facts from GD-IDS.yaml — pillars (the flow serves them), the systems in systems (a GOAL exercises a system through GOAL → SYS, refined to GOAL → AC), and terms already locked. These are constraints, not suggestions.
  • Exercised system GDDs — for each system this flow's GOALs touch, read its header and sections C (Detailed Design), H (Acceptance Criteria), and B (Player Fantasy) so each GOAL lands on a real rule/AC and the promised feeling is grounded. A GOAL aimed at a missing or deprecated system is a registry gap (route it through unikit-gd-spec add-system — Phase 4).
  • Recent unikit-gd-verify reports for this flow, if any (.unikit/gamedesign/reviews/).
  • Explore research (internal design lens) — if this flow was seeded from a unikit-gd-explore brief, discover the research deterministically (survives a /clear): read the flow's GD-IDS flows[].research: pointer (authoritative), falling back to the researches/INDEX.md entry whose Target: is this FLOW-<slug>. Read that research's RESEARCH_BRIEF.md → ## Flow Feature Plan block and use its objective/pacing/dependency seeds as the starting drafts — a seeded section is drafted silently in Generation (Phase 4) and confirmed in the group review (Phase 5); a seed is a draft, not an approved write. For a revision (Edit), the matching block is ## Flow Improvement Plan — its ready-to-apply delta lines, expected scale, touched GOALs/systems, and any RF-<date>-n it closes pre-fill the change set (the user still approves every edit). These explore-seeded drafts are untagged normal authored content (gd-provenance → Provenance) — the import-only extracted / generated markers do not apply to flows.

Phase 2 — Resolve Mode + Depth (no flags)

Check .unikit/gamedesign/flows/FLOW-<slug>.md and read the intent in the prompt:

  1. Does not exist → Create: build the skeleton (Phase 3), then run the Decision-First flow — the wiring mode is the first decision in Phase 4 (it shapes sections B and C).
  2. Exists and the prompt describes a change to approved content (a goal, pacing beat, dependency, event, or the wiring mode: "retune the pacing", "add a branch", "change a GOAL", "rework the onboarding") → Edit: classify the scale and apply the delta (Revision below). Approved text is changed only through this path.
  3. Exists with [To be designed] placeholders and the intent is to continue / fill (or no change is described) → Fill: resume from the remaining placeholders. Approved text is never overwritten — only placeholders are filled. Skip the skeleton sub-step (Phase 3.1); still fork-scan and run Phases 4–5 over the placeholders.
  4. Exists, complete, no change described → authoring is done; nothing to write. Offer review / verify (Handoff) and stop.

If the flow name is ambiguous (matches several entries, or none) → the question tool of your agent (AskUserQuestion) listing candidates. Never guess the target. Print the candidates to the screen as plain markdown first.

Depth gate (Create / Fill only). Once the mode is Create or Fill, present one depth picker — the named tiers core/standard/full, each with what it adds, not bare letters:

The question tool of your agent (`AskUserQuestion`): How deep should this pass go?
Options:
1. core (recommended) — the floor that gets the flow to a solid, usable base: Overview,
   Objective Flow (the run of objectives the player works through).
2. standard — core + Pacing, Dependencies, Events (the funnel).
3. full — every section, fully tuned.

The picked tier is the ephemeral scope of this pass — NOT stored (gd-authoring / gd-lifecycle); the lasting fact is the inferred status at Phase 6. Edit mode skips the depth gate entirely (it lies flat on the Revision flow).


Load the Mode Body

With the mode resolved (Phase 2) and the depth picked (Create / Fill only), load the matching body on demand and follow it — do not keep both bodies in context. Create / Fill returns to Phase 6 (Registry & State) below; Edit returns to Handoff.

Mode Reference body
Create / Fill {{skills_dir}}/{{self_name}}/references/mode-author.md — the Decision-First section-cycle: the A–F section map + core-set, Phase 3 (Skeleton + Fork-Scan), Phase 4 (Decision Interview + Generation, incl. the wiring-mode decision + the re-entry seam), Phase 5 (Group Review).
Edit {{skills_dir}}/{{self_name}}/references/mode-revise.md — classify the scale (Tuning / Tweak / Rework / Structural-redirect), apply the change, then the mandatory delta tail.

Phase 6 — Registry & State (write flows: / events:, then re-render the maps)

After the sections are authored and accepted:

  1. Write the GD-IDS.yaml flows[] entry (the flow's machine truth — the flow registers itself): id: FLOW-<slug>, name, status: active, mode, doc_status, version, depends_on (the SYS ids from section D), goals: (one GOAL-<flow>-<n> per row — id, summary, targets: the SYS/AC it exercises), research: (the explore-research folder if seeded — a non-id path; owned here, not by unikit-gd-spec), source: flows/FLOW-<slug>.md, added. Write each funnel event under events[] (id, name, status, source, flow:). Get approval before the write; existing values are never changed silently; every fact carries its source (gd-principles → Facts Registry).
  2. Status (inferred — gd-lifecycle). Set the flow's status in the two places that must agree — the FLOW.md header > Status: token and the GD-IDS.yaml doc_status: detailed once the whole core-set (A/B) is authored; held at skeleton while a core section carries <!-- deferred --> (the soft floor guard; the WARN in Phase 4); detailed · partial (n/m) (rendered, inferred from the markers) when the core is complete but ≥1 non-core section is deferred. Set version: 1 (header + GD-IDS). Write the initial changelog block to section F (#### v1 — <date> — initial design with the GOAL: + GOAL-<slug>-1 … N (new) line and Affected (gd-verify): —) — the single current block (K1).
  3. Regen-on-write — re-render the maps (B1). As the last step, re-render GAME.md's ## Flow Map [gen] and ## Funnel [gen] blocks from GD-IDS (the RULES_INDEX render model): replace only the content between <!-- gen:flow-map --> / <!-- /gen:flow-map --> and between <!-- gen:funnel --> / <!-- /gen:funnel --> — never touch the authored one-pager above, and never the ## System Map [gen] block (that is unikit-gd-spec's render).
    • Flow Map — one row per flows[] entry, grouped by wiring-mode (Linear/Conditional, Emergent): ID | Flow | Mode | Status | Ver | Depends (SYS) | Realized | Doc. Status mirrors doc_status (plus deprecated from the status field), with the · partial (n/m) suffix appended when the FLOW.md carries ≥1 <!-- deferred --> (the canonical render format from gd-lifecycle — the same suffix verify and the System Map use); Ver is — until skeleton.
    • Realized is DERIVED, never written: yes once every system in the flow's depends_on carries a non-empty implemented_version in GD-IDS (the code-set field), else no. It is computed from the systems' state on each render — there is no implemented field on flows.
    • Funnel — one row per events[] entry: Order | Event | Flow | Measures, ordered by funnel position.

Handoff & Next Steps

Recommend the next steps (do not auto-invoke):

  • ✅ Verify consistency & impact — /unikit-gd-verify FLOW-<slug>
  • 🔍 Qualitative review (fresh session) — /unikit-gd-review flows/FLOW-<slug>.md
  • 🧩 Detail an exercised system — /unikit-gd-system <SYS-slug>

A review is most independent in a fresh session (the reviewer should not have authored the doc). unikit-gd-verify is read-only — it checks the flow against the registry and prints dependent flows on cross-axis staleness; the ## Flow Map [gen] / ## Funnel [gen] were already re-rendered on this skill's write (B1).

Final: Compact Report

The terminal plaque (TIER A — gd-principles → Language: name leads, ids are parenthetical copy-paste tags, status is a plain phrase):

Flow: <name>  (FLOW-<slug>)
Mode: <linear | conditional | emergent>   ·   Depth: <core | standard | full>  (create/fill)
Action: <create | fill | edit (tuning|tweak|rework)>
Doc: .unikit/gamedesign/flows/FLOW-<slug>.md — <ready | ready, a couple of optional blocks left | skeleton, main blocks not filled yet>, vN
Registry: +<G> goals, +<E> events, depends_on [<name> (SYS-<slug>), …]  (GD-IDS.yaml)
Goals: GOAL-<slug>-1 … GOAL-<slug>-N   [GOAL delta on an edit: +<n> / changed <n> / removed <n>]
Maps: ## Flow Map [gen] + ## Funnel [gen] re-rendered

No summary document, no report file.

Ownership Boundaries

  • Owns: the full lifecycle of flows/FLOW-<slug>.md — creating the skeleton, filling placeholders, and revising approved content (Tuning / Tweak / Rework, with the version bump + changelog; the status stays detailed); the flow's GD-IDS.yaml facts (flows[] incl. goals, mode, depends_on, research:, and the flow's events) and its doc_status / version; the wiring-mode selection; and the re-render of the ## Flow Map [gen] and ## Funnel [gen] blocks in GAME.md (the [gen] regions only).
  • Not this skill: systems → unikit-gd-system; the system roster, the GAME.md one-pager content (pillars, win/lose intent, monetization stance) and its ## System Map [gen] render → unikit-gd-spec; quality verdicts → unikit-gd-review; consistency & cross-axis impact → unikit-gd-verify.
  • Never: write, fill, or edit without approval; overwrite approved text outside the delta discipline; skip the version bump or changelog on an edit (unrecorded delta); change a GD-IDS.yaml value silently; delete or renumber an ID; write the systems: roster or the GAME.md one-pager content; touch the ## System Map [gen] block; read the code workspace or project source.

Quick Reference

/unikit-gd-flow FLOW-first-session                      → select mode + create skeleton + author A–F (if no doc yet)
/unikit-gd-flow first-session                           → resolve to the FLOW-slug; same flow
/unikit-gd-flow FLOW-first-session                      → resume filling placeholders (partial doc)
/unikit-gd-flow FLOW-first-session "retune the pacing"  → Tuning: edit C/B, one approval
/unikit-gd-flow first-session "add a stealth branch"    → Rework: section-cycle old-vs-new
/unikit-gd-flow FLOW-onboarding "switch to conditional" → Rework: mode change (B + C restructured)

04 同仓库更多 Skills

NintendaDev
NintendaDev/unikit-ai

unikit

Initialize the UniKit framework in a project. Scans the current project to detect the engine and real tech stack, then generates the base framework documents that give AI agents context about the project — the resolved project description, the architecture doc, AGENTS.md, and the user-editable .unikit/config.yaml. This is the one-time, whole-project setup run before any other unikit skill. Use whenever the user wants to initialize, set up, or onboard the framework, e.g. "initialize unikit", "init the framework", "set up unikit", "set up the project", "configure AI context", "scan my project and set it up", "bootstrap the project". It also actualizes the config of a project already set up — say "update the config", "add missing settings", "actualize configuration", "fix my config". On a set-up project it offers the actions to choose from.

Skill与智能体开发
★ 22
NintendaDev
NintendaDev/unikit-ai

unikit-verify

Verify a completed implementation against the plan in .unikit/code/plans/. Checks that every task was implemented, nothing was forgotten, the code compiles, the recorded test run still matches the code, and {{engine_name}} conventions are followed. Use after /unikit-implement completes, or when the user says "verify", "check the work", "did we miss anything". For code review use /unikit-review instead.

代码审查Skill与智能体开发
★ 22
NintendaDev
NintendaDev/unikit-ai

unikit-todo

Manage the project's TODO list in .unikit/TODO.md — capture work to remember for later, not to do right now. Use it to add a deferred item, mark a todo complete, or show the list. The point is to park tasks you don't want to act on in the moment but don't want to forget. Trigger on "add a todo", "add this to the todo list", "put X on the todo list", "remind me to...", "don't forget to...", "note this for later", "we should do this later", "mark this todo done", "what's on the todo list", "what's left to do". This is only for parking reminders — to actually plan or do the work now, use /unikit-plan, /unikit-implement, or /unikit-execute.

★ 22
NintendaDev
NintendaDev/unikit-ai

unikit-skills-context

Customize how a built-in unikit skill behaves in this specific project by writing workflow overrides into its skill-context file. These are per-skill workflow rules (e.g. "/unikit-review should also flag TODO comments", "/unikit-implement should run tests after each task"), not coding conventions. Use when the user wants to change, customize, or override how a particular unikit skill works in this project — "customize the review skill", "make /unikit-fix always add logging here", "the implement skill should do X differently", "add a skill rule for /unikit-commit". Use "validate" to find stale overrides against updated base skills. For a general coding convention (not tied to one skill) use /unikit-rules; to derive rules from past mistakes use /unikit-evolve.

Skill与智能体开发
★ 22
NintendaDev
NintendaDev/unikit-ai

unikit-rules

Add a short, project-specific rule, convention, or override to .unikit/RULES.md — the quick-capture inbox for this project's rules (common rules always load; topic files in .unikit/rules/ load when the work matches), automatically loaded by /unikit-implement (later promotable via /unikit-memory migrate-rules). Works only with a rule typed as a prompt — it does NOT read files, folders, URLs, or PDFs. Cross-checks the knowledge base (RULES_INDEX.md) to avoid duplicating core/stack entries. Use for one-line conventions and corrections: "add a rule", "remember this", "convention", "always do X", "never use Y", "from now on do Z", or when the user corrects agent behavior and wants it remembered; also "split the rules into topics" (optimise), "clean up the rules" (prune), "shorten RULES.md" (compact), by keyword or in plain words. If the user points to a source (file, folder, URL, PDF, article, book) or wants to research/document framework usage, use /unikit-memory; for architecture decisions use ARCHITECTURE.md.

文档处理资料检索与研究
★ 22
NintendaDev
NintendaDev/unikit-ai

unikit-rules-registry

Orchestrate the remote rules registry — the shared registry (often connected as a local folder or git repo) that the project's rules are published to and pulled from. Three modes, picked from the user's wording: create scaffolds a new local rules registry ("create a local rules registry", "init a rules registry"); update migrates/promotes local knowledge-base rules from .unikit/memory/ into the remote registry with automatic semver bumps ("migrate local rules to the remote registry", "promote my rules to the registry", "push local rules", "publish local rules", "update the rules registry"); sync pulls registry updates back into memory ("sync the rules registry", "pull registry updates"). Pass "--module <id>" to scope to one knowledge-base module. This skill ONLY orchestrates the CLI (`unikit-ai rules *`) — it never edits .unikit.json or .unikit/memory/ sources directly.

★ 22