Menu
SkillsMP has collected 36 skills from davesnx/setup. Open a skill to review its source and details.
Showing 36 of 36 collected skills.
Design production modules, interfaces, seams, types, signatures, and dependency structure before implementation. Use ONLY when the requested result is a production code shape or integration contract, such as /architect, 'architect this', 'design this module',…
Review local changes, branch diffs, commits, or pull requests for correctness, security, regressions, maintainability, repository standards, specification compliance, and blast radius. Use for "review", "code review", "review my changes", "review PR", "review…
Use only when the user explicitly asks to model a domain, establish ubiquitous language, create or update a glossary or CONTEXT.md, map contexts, or create or revise an ADR. Do not use for TDD, test-first implementation, red-green-refactor, or integration…
Review, execute, reconcile, or publish an existing codebase audit plan. Use when the user points to an existing plan, plans directory, or audit backlog and explicitly asks to review the plan, implement it, check it against current code, reconcile its status…
GitHub and Git branch work: create validated commits, inspect PR state and review conversations, diagnose GitHub Actions failures, compose non-trivial gh commands, or resolve merge, rebase, and cherry-pick conflicts. Load before every `git commit`. Use for…
Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, says "grill me" or "grill with docs", or uses another grill trigger phrase.
Audit any codebase across correctness, security, performance, tests, architecture, dependencies, migrations, developer experience, documentation, and product direction. Render vetted opportunities as a local HTML review, prioritize them, then write selected…
Build throwaway, self-contained, drivable HTML artifacts to answer state-model, transition, policy, scheduling, or algorithm questions. Use ONLY when the requested result is a disposable interactive logic experiment used to test behavior before production…
Apply simplicity standards to all software engineering work. Use for every request to write, change, fix, refactor, review, simplify, or clean up code, including AI-generated code, reuse, quality, and efficiency work, and when choosing APIs, architecture,…
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or…
Use only when the user explicitly asks for TDD, test-first implementation, red-green-refactor, or integration tests as test-first work. Do not use for generic implementation, tests without a test-first request, bug fixes, or domain-modeling artifacts. If the…
Write, edit, or review developer documentation such as READMEs, RFCs, tutorials, how-to guides, reference pages, explanations, PR descriptions, runbooks, and architecture docs. Use when the requested result is developer-facing documentation, including…
Gather and assess technical evidence from primary sources, produce cited findings, and resolve or report source conflicts. Use when the main task is research, fact verification, source comparison, or a cited research artifact, even if the findings will later…
Move issues and external PRs through a state machine of triage roles — categorise, verify, grill if needed, and write agent-ready briefs.
Edit or review prose to remove AI-generated writing patterns. Use only for writing, documentation, copy, reports, or messages when the user asks to unslop or deslop text, remove AI tone, or clean up prose. Do not use for source code.
Write, edit, or analyze posts and threads for X, including audience fit, supplied analytics, and verified recommendation-system constraints. Use when the requested output or diagnosis concerns X content. Treat algorithm details as constraints and never…
Create branded architecture, IT current-state, flowchart, sequence, state machine, ER/data model, timeline, swimlane, quadrant, radar/spider, polar chart (polar/radial lollipop), loop/flywheel, nested, tree, org chart, layer stack, Venn, pyramid/funnel,…
Sync shared agent skills between the local Mac and nspawn through the setup Git repository. Use when the user says "sync skills", "push skills", "send skills to nspawn", "update nspawn skills", or asks to synchronize OpenCode, Claude, Codex, or Agent Skills…
Use when starting, configuring, or exposing a service inside systemd-nspawn.
Create a self-contained visual HTML explanation with concise diagrams and code-shape sketches. Use ONLY when the user explicitly asks to use the eli5 skill.
Use when the user wants to write, co-author, improve, or review a blog post or article. Trigger on a title, topic, outline, rough draft, complete post, 'write a post about,' 'improve this post,' 'blog draft,' or 'editorial feedback.' Uses an interview-first,…
Run a model-diverse subagent council to investigate the same problem from multiple perspectives, compare findings, and produce a final recommendation. Use this skill whenever the user asks for a council, second opinions, multiple agents/models to evaluate one…
Explain or research how something works in this codebase by exploring code and producing a clear architectural explanation. Use for "how does X work", "research X", "trace X", or "where is X used" when X refers to the current codebase. Optionally critique the…
Puts real design variations of one project file on the user's own localhost, behind a small card that flips them with the arrow keys, then keeps listening for their verdict. Use when the user wants design variations, alternatives, or directions to choose…
Writing, exploit: shape raw material into an article, paragraph by paragraph.
Browser automation CLI for tasks that require interaction with a rendered website, such as navigation, forms, screenshots, scraping, login, or web-app testing. Do not use for GitHub pull requests, commits, or repository files when git, gh, or code-review can…
Diagnosis loop for hard bugs and performance regressions. Use when the user explicitly asks to diagnose or debug, or provides a concrete runtime failure or measured regression. Do not use to judge whether a pull request or commit is the wrong solution; use…
Use for frontend visual design and UX work, including layout, styling, accessibility, responsive behavior, themes, typography, color, motion, interaction, UX copy, and design systems. Not for React implementation or performance work alone; use…
Reconstruct recent working context from OpenCode, Cursor, or other available chat history, live repository state, and shared records such as issues, prior fixes, and incidents. Use when the user says 'recall my work on X', 'find our conversation about X',…
Use for explicit React or Next.js performance work involving rendering, data fetching, bundles, RSC, hydration, or measured regressions. Do not use for visual design, layout, styling, accessibility, or general component edits; use impeccable for those. Load…
Set up and run a bounded, measured experiment loop for an optimization target. Use when asked to start autoresearch, optimize a metric, or run experiments.
Inspect, plan, fix, and monitor failing CI across GitHub Actions, Buildkite, and other attached providers. Use when a user asks to debug CI, summarize a red build or PR, propose a CI fix plan, fix failing checks, retry a flaky job, or watch checks until they…
Understand a branch or make its pull request easy to review without changing behavior. Use for "prepare branch context", "what's on this branch", "summarize this branch", "prepare PR", "make this easy to review", "tidy this PR", "clean up commits", or…
Spawn three parallel review subagents over the active transcript, surface learnings, and route each to a concrete edit on an existing skill. Use when the user says reflect.
Turn the current conversation into a spec and publish it to the project issue tracker — no interview, just synthesis of what you've already discussed.
Break a plan, spec, or the current conversation into a set of tracer-bullet tickets, each declaring its blocking edges, published to the configured tracker — edges as text in one file per ticket locally, or native blocking links on a real tracker.