Claude Skills by OutlineDriven

Use when a user asks for a tech spec or architecture doc. Produces a self-contained TECH.md that translates intent into an executable implementation plan. Not for PRDs — use write-prd; not for product specs with behavioral invariants — use write-product-spec.

Use when asked to push an already-working artifact to finished composition, texture, and feel without confusing implementation completion with composition. Not for remote, credential, publish, deploy, or irreversible changes. Not for hygiene routing — use clean-and-true.

Use when asked to build spring, easing, gesture, and choreographed animations with correct physics and reduced-motion support. Not for auditing existing motion or finding animation opportunities.

Use when a brainstorming question is genuinely clearer shown than told: a real mockup, layout, or diagram. Starts a local key-gated HTTP/WebSocket server, pushes HTML fragments to the browser, and merges click selections back into the terminal reply. Not for text-only idea exploration; use brainstorm or ideate.

Use when the user requests an accessibility audit, a11y check, or WCAG compliance review. Returns a prioritized WCAG findings report with file and line locations, before and after code fixes, and manual testing recommendations. Don''t use for tasks that require source or remote-system changes.

Use when /design-consultation proposes a design system with mockups or HTML and persists the approved system in DESIGN.md. Not for in-chat direction picking — use design. No remote, credential, publish, deploy, or irreversible changes.

Use when creative work is requested and no approved design exists, or when a raw idea or repository must be developed into an approved design document. Routes through idea, spike, bounded, or architectural design and fires before any implementation action. Not for work with an approved design already in context. No remote or irreversible changes.

Use when /design-variants generates distinct design directions and a gallery for selection. Not for one polished artifact — use polished-web-prototype; not for token-system direction picking — use design. No remote, credential, publish, deploy, or irreversible changes.

Use when a user builds or styles a web frontend or asks to make it not look AI-generated. Commits a token system and crafted interface passing a slop-audit and WCAG 2.2 AA gate. Not for component-level UI work without the token system; use frontend-ui-engineering. Local writes only.

Use when asked to build or modify user-facing interfaces and pages with production quality: components, layouts, state management, WCAG 2.1 AA accessibility, and real content, with output that looks hand-crafted rather than AI-generated. Not for token-system design with slop-audit gates; use frontend-design-deslop. Local writes only.

Use when the user invokes /ios-visual-review to audit an iOS app''s visuals on a real device. Produces a per-screen visual-review report scoring ten dimensions 0-10 with a biggest-leverage fix for each. Don''t use for remote, credential, publish, deploy, or irreversible changes.

Use when the user runs /polished-web-prototype or asks to turn an approved mockup, design plan, or freeform description into a polished responsive Pretext-native HTML artifact refined through a live preview loop. Writes the artifact and finalized.json under the project design directory. Not for variant galleries; use design-variants. Not for live-URL audits; use web-design-review.

Use when deciding what an interface should do before UI is built or audited: interaction consequences, action scope, reachable states, naming, and accessibility of the product decision. Read-only analysis; produces a chat report with findings and routed follow-on work. Not for visual implementation; use prototype.

Use when writing, editing, cleaning, or refactoring Tailwind classes, components, or configuration. Reorders and deduplicates classes, replaces arbitrary values with project tokens, and checks component extraction. Not for read-only audits of Tailwind code, or general CSS without Tailwind.

Use when asked to audit typography across a codebase. Produces a file:line report with concrete CSS and HTML fixes ordered by impact across 10 categories, and records rules that need rendered metrics as not-verifiable-from-source instead of fabricating violations. Not for building a type system or token scale.

Use when directing, building, or auditing React/Next.js UI in Tailwind: visual direction, responsive or dark-mode retrofits, and file:line UX findings with a ship verdict. Not for deep typography or motion passes; use typography-audit or ui-animation.

Use when the user runs /web-design-review with a URL to visually audit a live UI through the browser and fix findings in a screenshot-verified loop. Not for design direction picking; use design. Not for variant galleries; use design-variants.

Use when asked to design, write, review, refactor, lint, or performance-diagnose Lean 4 proofs, libraries, or tactic extensions under Mathlib conventions. Not for non-Lean code or changes outside Lean source, library API, proof structure, and linter configuration.

Use when the user says "audit my code", "find all the bugs", "review until clean", or "grill my changes". Runs an iterative multi-agent review that resolves every finding at or above a configurable severity floor to zero, or stops at a gate or cap. Not for remote, credential, or irreversible changes.

Use when product copy needs buyer-objection evidence collected through approved outreach. Conducts structured interviews via approved channels with consent, then synthesizes a copy recommendation grounded in anonymized exact language. Not for unsolicited outreach or survey design.

Use when a measured cost surface needs one-change-at-a-time reduction under frozen guardrails. Each proposed change is adopted with N=5 gate evidence or recorded dead with the measurement that killed it.

Use when a long agentic project needs each cycle to produce a learning memo with lessons, a keep/iterate/restart decision, and version labels only on quality-cleared artifacts. Drives each slice through the real surface and captures evidence before iterating. Not for single-pass builds.

Use when environment-dependent code, templates, or deployment configuration changes, or when runtime configuration is missing. Produces a bidirectionally reconciled environment-variable contract.

Use when recent resolved feedback may reveal a broader recurring defect pattern: sweep resolved feedback for shared root causes, search the project surface for generalized recurrence, and verify no recurrence remains. Not for source-level feedback collection; use feedback-sweep for that.

Use when a first screen needs lower transfer bytes without visual or behavioral change: measure compressed bytes, reduce assets and bundles to a budget, prove pixel identity, and pass the test suite. Not for visual redesign or behavioral changes.

Use when a visual needs repeatable fixed-view rendering and independent rubric scoring: render the fixed view through a specified interface, score it against a frozen rubric, and prove the saved render clears the threshold. Not for free-form visual review or subjective critique.

Use when a rendering path needs stable frame-time, CPU, GPU, and memory evidence against fixed targets: stabilize the configuration, define a sampling window, and prove every target with two consecutive comparable runs. Not for one-shot profiling or visual quality review.

Use when an authorized reference surface needs a clean-room frontend reconstruction across static, motion, and responsive fidelity. Captures the reference at declared viewports and timelines, reconstructs, and measures fidelity against tolerances. Not for styling or component work without a reference surface.

Use when a complete product, rather than one component, needs production-like acceptance evidence against its documented acceptance criteria. Traverses the capability inventory, records per-capability evidence, and stops at declared success, non-success, or bound. Not for single-component evaluation or evaluation without documented criteria.

Use when a genealogical identity or relationship needs correlation, conflict analysis, and negative-search proof. Applies the Genealogical Proof Standard to classify the proposition as proved, disproved, likely, or possible. Not for non-genealogical identity proof.

Use when a test surface needs coverage raised to a configured target with behavior-guarding assertions, not vacuous execution-only tests. Binds coverage metric and tool, mutation tool and kill threshold, measures baseline, adds guarding tests, and re-measures. Not for line-coverage inflation without mutation proof.

Use when agent strategies need a reproducible finite tournament under a frozen evaluation protocol. Produces a scorecard with move records, scores, and hidden-identity validation. Not for open-ended or infinite matchups — use a continuous benchmark for those.

Use when one named review viewpoint must run fix cycles until a fresh reviewer finds nothing. Leaves verified machine-owned fixes and a written verdict. Not for multi-viewpoint review or remote, credential, publish, deploy, or irreversible changes.

Use when a user asks to iteratively improve a skill, fix skill-quality findings, or resolve scanner alerts (W001, W011, W012) on a skill. Resolves the target, runs structured reviews or scanner-backed remediation, applies fixes with per-edit re-scan, and repeats until clean. Not for agent grading; use skill-doctor for that.

Use when a bug, failure, flake, regression, review finding, or ticket needs the core fixed so it cannot break the same way again. Reproduces the failure, proves the root cause, repairs at the root, and pins with a regression test. Not for greenfield features — use tdd; not for style-only review or typo-class one-liners.

Use when a test suite is too slow and must be accelerated without weakening behavior or coverage. Also handles CI parallelization when the bottleneck is serial execution. Not for deleting tests — use tests-purge-unneeded; not for remote, credential, publish, deploy, or irreversible changes.

Use when thumbnail concepts need real-size, accuracy-first scoring without misleading claims. Renders at real platform dimensions, scores against a fixed rubric with weighted accuracy dimensions, and writes an immutable receipt. Not for generating thumbnails or declaring winners that fail the accuracy rubric.

Use when a human explicitly invokes a watcher cycle for an open pull request that must be watched until mergeable or blocked. Classifies the PR and addresses bounded review findings on its own branch. Don''t use for merge, force-push, or work outside the invoked cycle.

Use when authorized workflow history may contain a repeated process worth extracting and replay-testing. Produces a contradiction-tested workflow mined from authorized history.

Use when inspecting, searching, verifying, or parsing DWARF debug info: DIEs, DW_TAG_/DW_AT_ entries, .debug_* sections, line tables, or llvm-dwarfdump/readelf output. Returns the requested query, verification, explanation, or parser with tool and platform identified.

Use when C++ code is being written or reviewed and the compiler supports modern safe idioms. Returns a structured guidance report on supported features, anti-pattern replacements, and compiler-compatible hardening flags. Read-only; no source or remote-system changes.

Use when the user runs /autoplan on a plan or idea. Runs CEO, design, DX, and engineering review phases over the idea, amends the plan, derives deterministic task IDs into a task JSONL ledger, and waits at a final human approval gate. Not for remote, credential, publish, deploy, or irreversible changes.

Use when asked to park an undecided idea without representing it as decided or active work. Creates or updates one reversible local backlog entry with evidence and a readiness threshold. Not for decided or active work — use the project task system.

Use when the user faces multi-fork decisions or unresolved prerequisites. Maps a decision tree and asks its frontier as batched single-select questions until shared understanding is confirmed. Also handles "batch questions". Not for one-at-a-time interviews — use interview-me.

Use when the user begins knowledge work with notes, a transcript, a brain dump, or a problem. Captures original wording, reads prior knowledge, resolves up to three load-bearing questions, and writes a brainstorm file before planning. For divergent option generation, use ideate.

Use when the user says "clarify", "what is ambiguous here", or when a prompt or spec reads as under-specified before planning. Surfaces ambiguities as a certainty-tiered findings report with recommended defaults. Not for exploring intent with questions — use askme.

Use when the user has a fork and wants it resolved and applied, not explored: "decide this", "choose the path", or "decide and fix it". Grounds the fork, frames one single-select with a recommended default, takes the pick, and applies it in the same turn.

Use when asked to derive the general rule a request carries when it arrives as examples instead of a stated rule, then bound that rule. Not for ambiguity resolution inside a stated request — use clarify. Read-only.

Use when a plan path or text is supplied. Audits every item in the plan against the current codebase, including targetless decisions and assumptions, and produces an HTML accuracy verdict page. Not for tuning review questions, use plan-review-tune; not for scoring, use planning.

Use when asked to judge whether this project should adopt, switch to, reject, or revisit a technology, library, pattern, or architecture, or give a mid-session second opinion. Graded verdict clears two evidence floors. Not for scoping, use brainstorm; not for forks, use decide.