alfonsograziano

Generate a PLAN.md implementation plan from a SPEC.md file. Use when the user wants to plan implementation for a spec, says "[P]" after creating a spec, or asks to create an implementation plan. The plan translates the spec's "what" into a technical "how" — approach, trade-offs, sequencing, and a task checklist — without writing any code. Trigger whenever someone says "let's plan this", "create a plan for", "plan the implementation", or provides a spec path and wants to move to the planning phase.

Implement a spec from a specs/ folder, following the PLAN.md task list if present. Use when the user wants to start coding, says "[I]" after reviewing a spec or plan, provides a spec folder name and wants implementation, or says things like "let's implement this", "start coding", "implement the plan", or "build this". Always trigger when there's a spec folder path and implementation intent, even if the user just says "[I]" or "go".

Create or update a CONSTITUTION.md file in the specs/ folder by scanning the project. The constitution captures project-wide context that all other nano-spec skills rely on, like tech stack, sanity check commands, coding conventions, key files, and hard rules. Use when the user wants to set up nano-spec for a new project, initialize the constitution, or keep it up to date. Trigger when the user says things like "create the constitution", "set up the project constitution", "initialize nano-spec", or "update the constitution".

Create a SPEC.md file for a new feature, product, or system using the Spec-Driven Development (SDD) approach. The spec works in the problem space — it clarifies the "what", not the "how". Use this skill whenever the user wants to write a spec, define requirements, capture what needs to be built, create a specification document, or start the SDD workflow. Trigger even if the user says things like "let's spec this out", "I want to define what we're building", "help me think through requirements", or "let's write a spec for X".

Generate a client-ready PowerPoint deck from the filesystem template library in this repo. Use when the user gives context, a brief, a narrative, notes, or raw content and asks to create, build, compile, regenerate, or QA a .pptx deck from the imported slide templates. Also use when asked to choose slide templates, write a deterministic build.ts script, run the local generator, inspect rendered screenshots, fix build failures, or return the final PPTX and report.

Customize the deck design system (colours, fonts, slide grid, logos) so generated decks match a specific brand. Use when the user wants to change the palette, set brand colours, swap fonts, add a logo, adjust the slide layout, or "make it look like our brand". Edits design.md and src/design.ts together, and optionally installs the new fonts. Trigger on requests like "use our brand colours", "change the accent to purple", "set the font to X", "add our logo", or "customise the design".

Draft the description.md for a slide template in templates/ by looking at its rendered screenshot (or its fields when no screenshot exists). Use whenever the user wants to describe, document, or write up what a slide template looks like and when to use it, fill in a template's TODO description stub, or asks "what is this slide for" about a template in the library. Triggers on a template name, or a request to document one or more slide templates. If no name is given, ask for it; if several are given, do each one.

Ingest one or more slides from a PowerPoint deck into the pptx-gen template library and document each one. Use when the user gives a source .pptx file (often a single slide or a deck downloaded from Google Slides) plus a list of slide numbers and a template name, and asks to import, ingest, or add slides as reusable templates in templates/. Runs the ingest CLI for each slide, then runs the draft-slide-template-description workflow on each imported template so it ships with a real description instead of a TODO stub.

Generate an HTML accuracy chart for an auto-agent job. Use this skill whenever the user asks to create, generate, or build an accuracy chart, performance chart, or iteration visualization for a job folder. Also trigger when the user mentions charting agent accuracy, visualizing hypothesis iterations, or creating a report for a job run — even if they just say "make a chart for agent-3" or "generate the HTML for this job". The input is a job folder path (e.g., jobs/agent-1/).

Rules and frameworks for writing high-quality LLM system prompts. Use this skill whenever you are authoring, reviewing, editing, or optimizing a system prompt — including when generating a system prompt as part of building an agent, tool, or workflow. Trigger on any task that involves writing instructions for an LLM, crafting a system message, tuning prompt wording, or reviewing an existing prompt for quality. Even if the user doesn't say 'system prompt' explicitly, use this skill when the output will be consumed as LLM instructions.