djalmajr/skills
31 skills1.9k installs93 starsGitHub
Install
npx skills add https://github.com/djalmajr/skillsSkills in this repo
1Agile Tddagile-tdd guides a test-driven-development workflow and overall testing strategy, coaching red-green-refactor, writing new tests, and establishing coverage for a module. A solo builder reaches for it when starting a feature test-first or when adding tests to existing code. It gives structure to disciplined testing rather than ad-hoc test writing.91installs2Agile Epicagile-epic turns an intake, roadmap, or large initiative into a structured epic: an overview plus individual story files with tasks, dependencies, a roadmap diagram, and acceptance criteria. A solo builder reaches for it when work spans several coordinated stories or has cross-delivery dependencies that need sequencing. It chains into agile-story for detailing each item.89installs3Agile Roadmapagile-roadmap transforms broad objectives into a clear, prioritized roadmap that sequences multiple phases and surfaces dependencies, then connects it to the executable backlog. A solo builder reaches for it when work spans several phases, when early decisions constrain later ones, or when stakeholders need to see the whole journey at once. It scales from short multi-phase initiatives to quarterly roadmaps.86installs4Agile Sprintagile-sprint plans a work cycle by declaring the sprint objective, selecting backlog items based on value, risk and dependency, recording capacity and constraints, and setting an execution order. A solo builder reaches for it at the start of each cycle to align what will actually be done. It sits between backlog planning and day-to-day status tracking.86installs5Agile Storyagile-story creates a clear, ready-to-implement execution plan with tasks for a single story or a small, localized change that fits within one cycle. A solo builder uses it when a piece of work is already scoped by an epic, or is small enough not to need decomposition, and just needs a concrete plan before implementation. It writes the artifact in the user's language.86installs6Agile Protoagile-proto builds standalone, interactive UI prototypes using a CDN-only stack (z-proto, Tailwind v4, shadcn-style components, Preact/htm) so a solo builder can validate screens and flows before committing to real implementation. It can produce faithful send-to-Figma captures and stores prototypes alongside the initiative's planning artifacts. Reach for it when you want to mock up or explore an interactive UI flow fast.85installs7Agile Retroagile-retro conducts a retrospective that turns reflection on a finished cycle, sprint, or delivery into concrete, trackable improvement actions, using facilitation styles such as Start-Stop-Continue, Mad-Sad-Glad, or 4Ls. A solo builder or team reaches for it after a delivery to capture what worked and what to change. Scope can be per-story, per-sprint, or per-epic.85installs8Agile Reviewagile-review consolidates a sprint's deliveries into a clear review and demo format for stakeholders, covering what was delivered, what changed in scope, what is pending, and what decisions are needed. A solo builder or team reaches for it at the end of a cycle to present progress crisply. It tailors depth and content to the audience, from team-only to external stakeholders.84installs9Agile Skill Feedbackagile-skill-feedback captures what actually happened when an agile skill was used and turns it into evidence-backed proposals to refine, merge, split, deprecate, remove, or create skills and templates. A solo builder maintaining a skill library reaches for it when a skill did not fit a real workflow or when skills overlap. It keeps changes small, traceable, and validated against realistic artifacts.84installs10Agile Routeragile-router is a dispatcher that inspects your context and points you at the right agile skill, such as choosing roadmap versus epic or which ceremony or tracking skill fits next. A solo builder reaches for it when they know they need the agile flow but are unsure which artifact to produce. It is the navigation layer over the rest of the agile skill set.82installs11Agile Metricsagile-metrics is a Claude Code skill that consolidates objective sprint metrics such as velocity, lead time, cycle time, defect rate, and coverage into a quantitative summary. A solo builder uses it after a cycle to get hard delivery data that feeds retrospectives, sprint reviews, and capacity planning rather than relying on intuition.81installs12Agile Onboardingagile-onboarding walks a new team member through the agile-plus-AI operating model and artifact flow in a practical, progressive way, tailored to their role and desired depth. A solo builder or small team reaches for it when someone joins or changes roles and needs to become productive fast. It emphasizes hands-on practice over reading documentation.81installs13Agile Intakeagile-intake transforms vague problems, raw ideas, and unstructured requests into a clear, actionable intake document, then recommends the next artifact in the flow (roadmap, epic, or story). A solo builder reaches for it at the very start, when a need is not yet scoped enough to plan or build. It is the entry point of a larger agile-plus-AI delivery skill set.80installs14Agile Refinementagile-refinement operates in two modes: it lints planning artifacts for cross-references, dependencies, and format, or it reviews changed code for security, coherence, scope, and quality. A solo builder reaches for it as a quality gate before shipping, to catch issues in either the plan or the diff. Depth ranges from a quick lint to a full or security-focused review.80installs15Agile Statusagile-status tracks delivery progress against a plan, story, or epic in three modes: a daily checkpoint, a period consolidation report, and a formal closure, and it makes blockers explicit with impact, owner, and next action. A solo builder reaches for it to keep an honest running picture of where a delivery stands and to close it out cleanly. It complements sprint planning by covering execution and wrap-up.80installs16Aim Initaim-init wires a repo into the ai-memory system so agent sessions auto-capture and recall durable knowledge: it sets up the .ai-memory.toml routing marker, the recall/write snippet in CLAUDE.md or AGENTS.md, the MCP server entry, and the capture hooks. A solo builder reaches for it to give their coding agent persistent memory, greenfield or brownfield, including migrating off the old qmd/wiki stack. It also ships doctor, refresh, and multi-user flows for diagnosing and upgrading an existing setup.59installs17Figma Capturefigma-capture is a Claude Code skill that captures a rendered local web page or app screen and puts it on the clipboard for pasting into Figma, using the official html-to-design capture.js flow. A solo builder reaches for it to copy the current browser page into Figma or to add a dev-only capture button, so live UI becomes an editable design artifact without manual re-drawing.56installs18Aim Writeaim-write is a Claude Code skill that persists a durable page to an ai-memory MCP instance for decisions, rules, gotchas, schemas, or operational constraints the user explicitly wants remembered. A solo builder reaches for it on 'remember this' or 'save this decision', using path conventions like decisions/ADR-*, _rules/*, and gotchas/* so the right kind and highest-signal tier are recorded cleanly.45installs19Aim Queryaim-query is a Claude Code skill that explicitly recalls durable knowledge from an ai-memory MCP instance, using semantic/hybrid search, recent-page listing, or an open-ended prose digest. A solo builder reaches for it to ask 'have we decided X', 'what's the state of Y', or 'catch me up' before proposing a design, pulling in shared _rules and cross-project scopes that auto-recall would miss.44installs20Excalidrawexcalidraw is a Claude Code skill for authoring and editing .excalidraw scene files - the flat JSON model of rectangles, text, and arrows on an infinite canvas, including two-way arrow binding, z-order, and groups. A solo builder reaches for it to create or update diagrams directly, either hand-editing small scenes or building them programmatically with Excalidraw's own compiler, an SVG preview, and a geometry linter, with Mermaid supported as an input format.37installs21Workwork is a Claude Code orchestrator skill that makes a whole quality-pattern system automatic: it names the task shape, then picks the phases, patterns, and sub-agent mechanics to run, delegating to companions like work-check, work-refute, work-judge, work-exhaust, and work-gaps. A solo builder reaches for it to run a rigorous, self-verifying review-and-harden pass over changes instead of orchestrating those sub-agents by hand.35installs22Work Checkwork-check is a Claude Code self-verification skill that spawns a verifier sub-agent, parses its schema-validated verdict, and loops on fixes until the work passes. It is the default code verifier the work orchestrator runs before refutation, and a solo builder reaches for it to prove a change is correct rather than assuming it.35installs23Work Exhaustwork-exhaust is a Claude Code skill that implements the loop-until-dry quality pattern: it runs repeated rounds of discovery until no new findings surface, sizing an unknown problem by exhausting it. A solo builder reaches for it when a search, audit, or sweep must be complete rather than sampled.35installs24Work Gapswork-gaps is a Claude Code skill implementing the completeness-critic pattern: after find, dedup, and verify, it asks what the list or scope itself never covered and emits a machine-checkable clearedToClose verdict. A solo builder reaches for it as the final gate to catch omissions before declaring work done.35installs25Work Judgework-judge is a Claude Code skill implementing the judge-panel quality pattern: it explores a wide solution space by generating diverse candidates in parallel, judging them on evidence, and synthesizing a winner that grafts in the best of the losers. A solo builder reaches for it when a design or solution choice deserves several real options weighed against each other rather than the first idea.35installs26Work Refutework-refute is a Claude Code skill implementing the adversarial-verify quality pattern - a refutation tournament that runs evidence-weighted, perspective-diverse lenses over each claim to destroy weak conclusions. As the destructive verify stage of the work harness, a solo builder reaches for it to stress-test findings and decisions before trusting them.35installs27Work Reviewwork-review is an invoke-by-name Claude Code skill that runs an extremely strict maintainability review of the current git diff, checking abstraction quality, giant files, duplication, layering leaks, and spaghetti conditionals. It applies diverse review lenses, measures with oracles before opining, and sweeps outside the diff for completeness. A solo builder reaches for it when they want a harsh, ambitious code-quality audit before merging rather than a light local cleanup.35installs28Work Sweepwork-sweep is a Claude Code orchestration skill that runs several blind, independent searches over a codebase from different angles - by structure, semantics, git history, entity, and error paths - keeping each searcher ignorant of the others, then merges results at a barrier that dedups against a persistent SEEN set and lists what no angle covered. A solo builder reaches for it when one search strategy is known to be incomplete, such as 'find all X', a security review, or tracing where something can break.35installs29Work Tournamentwork-tournament is a Claude Code skill that implements the same task multiple ways in parallel isolated git worktrees, each from a distinct angle, gates every candidate through a build, typecheck, test, and lint oracle, then grades the survivors from their actual diffs rather than their self-reports, applies the winner to main, and re-verifies the oracle green. A solo builder reaches for it to run a best-of-N on a tricky implementation where picking by outcome beats picking by pitch.35installs30Ux Flowsux-flows is the catalog and orchestrator for a web application's end-to-end usage flows: it keeps the project's flow definitions and hands them one at a time to the ux-persona skill to validate usability in the browser. It separates the what (the flows, as project data) from the how (the usability walkthrough), and is stack-agnostic. A solo builder reaches for it to list flows, create new ones, or run a full usability walkthrough with replayable evidence.30installs31Ux Personaux-persona has the agent embody a user persona and walk through a single usage flow entirely through the UI, starting from the real entry point rather than internal URLs, to judge whether the journey is actually usable. It records actionable findings with Webwright-style replayable screenshots and logs. A solo builder reaches for it to test as a user, check whether a screen or journey is usable, and catch UX problems that passing code hides.30installs