
Use when the user asks to commit everything, smart commit, group commits, organize staged or unstaged changes, create logical commits, push current work, or preserve a clean commit stack from an existing working tree.
Use when starting or resuming any Khuym project session, choosing the next Khuym skill, running go mode, checking onboarding/scout state, or enforcing workflow gates. Bootstrap meta-skill for routing across the Khuym agentic development ecosystem.
Use when the user wants to turn a rough idea, vague task, feature wish, bug-fix intent, or Codex /goal objective into a clear verifiable goal prompt. Use when the user mentions goal mode, /goal, set_goal, long-running autonomous Codex work, or asks to be interviewed/grilled before creating a goal.
Use when Codex should turn a concept, topic, tutorial, lesson, algorithm, system, workflow, or abstract idea into a playable browser game that teaches through interaction. Use for planning or implementing educational web games with the Game Studio plugin, including concept-to-mechanic translation, gameplay loop design, learning progression, and browser-game QA.
Create self-contained interactive HTML visual explainers that turn difficult topics, concepts, code, math, systems, or arguments into memorable explorable artifacts. Use when Codex should research a topic, choose an interaction model, and build an artifact with simulations, diagrams, charts, sliders, hover inspection, scrollytelling, or small game mechanics so readers can explore and understand the message.
Codebase intelligence support skill for Khuym using the gkg MCP tools. Use when planning or discovery needs an architecture snapshot, file/definition discovery, existing-pattern evidence, importer lookups, or a quick symbol trace in a supported repo. Primary path: scout readiness with `node .codex/khuym_status.mjs --json`, then `repo_map` plus `search_codebase_definitions` plus `read_definitions`.
Use when the user asks to fine-tune on books, create an SFT dataset from books, train a style-transfer or author-voice model, extract ePub text, segment long-form book content, or prepare literary data for LoRA or small-model training.
Strengthen a raw user prompt into an execution-ready instruction set for Codex or another AI agent. Use when the user wants to improve an existing prompt, build a reusable prompting framework, wrap the current request with better structure, add clearer tool rules, or create a hook that upgrades prompts before execution.
Research-first feature discovery for unfamiliar, ambiguous, or high-risk implementation work. Use when Codex should map the real repo stack, find reusable local code, check upstream patterns, and verify current official docs before planning or implementing a feature.
Use when creating a new khuym skill, editing an existing khuym skill, or verifying a skill works under pressure before deploying. Use when you need an agent skill to be bulletproof against rationalization. Do NOT use for project-specific AGENTS.md conventions or one-off solutions.
Use when the user asks to create a persistent repo-local Codex agent, scaffold `.codex/agents/*.toml`, define a reusable specialist role, or match an agent to existing local skills and MCP servers.
Use when the final swarm phase completes. Run specialist review, artifact verification, UAT, review beads, finishing checks, and handoff to compounding.
Use when running inside a swarming worker. Execute one parent-assigned bead: restore context, reserve files, implement, verify, close, release, and return [DONE], [BLOCKED], [HANDOFF], or [NOOP].
Use when blocked workers, test failures, build errors, runtime crashes, or integration issues need systematic root-cause debugging. Reads history/learnings/critical-patterns.md to avoid re-solving known issues. Writes debug notes that compounding can later capture.
Use when exploring has locked CONTEXT.md and the work needs planning before validation.
Use when planning has an approved work shape and needs feasibility validation before swarming.
Use when reviewing completes or work is intentionally abandoned. Extract durable patterns, decisions, and failures into history/learnings, then promote only critical reusable lessons to critical-patterns.md.
Use when a feature request needs user-facing decisions captured in history/<feature>/CONTEXT.md before Khuym planning. Clarifies fuzzy scope without implementation research or bead creation.
Use when validating approves execution. Orchestrates bounded Codex workers, local reservations, worker results, rescues, and phase handoff. Tends the swarm but never implements beads directly.
Use when designing or implementing a premium animated landing page with AI-generated visuals or looping video, especially when the workflow involves Higgfield-style image and video generation, Google AI Studio-style code prompting, React or Tailwind landing-page implementation, dark-mode video blending, scroll-tied playback, aspect-ratio-matched media blocks, or motion-first section design.
Refresh README files and project documentation so they match the current repository state. Use when Codex needs to update stale docs after feature work, sync documentation to current commands, options, configuration, and workflows, rewrite changelog-style wording into evergreen current-state documentation, or turn a rough docs-update prompt into an execution-ready instruction set.
Use when you need a manual dream-style consolidation pass over Codex artifacts and existing Khuym learnings, including bootstrap-first scans, recurring-window updates, ambiguity resolution for merge/create new/skip, and approval-gated critical-pattern proposals.
Bootstrap a new AI-agent session by reading the repository operating docs and rebuilding project understanding from source. Use when Codex needs to start a new conversation, get up to speed on an unfamiliar repo, read AGENTS.md and README.md completely before acting, investigate the codebase to understand the project's purpose and architecture, or turn a rough onboarding prompt into an execution-ready repo-orientation prompt.