agent-workflow

An agentic skills framework for AI coding assistants - composable workflows for planning, TDD, debugging, and code review

20 skills · 3 commands · 1 agent

Writing-craft toolkit for coding agents: poems and song lyrics, game dialogue and barks, instructions and docs, textbook teaching, reader targeting and single or batch passage suitability assessment, academic and professional prose, screen and stage scripts, YouTube scripts, speeches, professional presentations and webinars. Twenty-one skills include prose-reading-level for evidence-linked rubric judgments, proposed changes and saved suggestion dispositions, and prose-annotate and prose-annotation-review for source-bound highlights, vocabulary, explanations, links, a shared review workbench and portable publication data for Markdown, static HTML and structured content. Native drafts are measured and linted against cited craft rules. Illustrated storyboards retain editable panels, choices and character quest tracks. Variant sets, owner reviews and a taste model support rewrites; prose audit reports style hallmarks, never authorship. Slide-file creation uses the available presentation tool.

21 skills

Skills for working in Unreal Engine 5.8 from a coding agent: choose the engine system that fits a goal, with a reference per system (Lumen, Nanite, Virtual Shadow Maps, Niagara, Gameplay Ability System, Enhanced Input, AI, Sequencer, networking and more); audit a project's structure, tests and performance practice against Epic's documentation and plan the fixes; build, package and profile a native C++ game; import sprites, meshes and sounds with a guarded importer; and author for UEFN. Includes example scripts that list plugin states, dump the Python surface and build and validate Niagara scenes.

15 skills

Editable pixel-art authoring for agents: reusable parts, independent child animation, exact frame timing, construction replay, offline normal-map lighting, paired atlases, GIF proofs and a shared live workbench.

14 skills

Evidence-led marketing for family-made tools, apps and games: audience discovery, channel strategy, paying-demand and non-English market research with sourced decision reports, campaigns, collateral, monitoring, reception and experiment decisions, and game-launch planning with readiness gates and weekly maker-hours checks, with decisions, next steps and collateral picks tracked in the ledger and the desk. Ten skills, a portable evidence ledger, a dependency-free Node CLI and a local campaign desk.

10 skills

Skills for building, debugging and proving Phaser 4 games from a coding agent, with browser probes and a test harness behind the claims; pairs with agent-sprites, agent-meshes and agent-beeps assets.

10 skills