System / public projection
The sum of what you save.
Sources become signals. Stable owners change only through named loops, versioned workflows, exact authority, and proof.
Goals
6
measurable promises
Workflows
14/16
ready command contracts
Owners
192
stable write targets
Reviews
3295/3376
final source decisions
ObjectsKnowledge / change view
04 / canonical owners
A small operating graph, not a page explosion
Signals accumulate on stable owners. Search these objects to see where agents should write durable change.
architecture:brain-systemKevin Brain SystemCanonical public architecture for evidence, objects, workflows, agent interfaces, and proof.knowledgearchitecture:source-fabricBrain Source FabricSource identity, artifacts, revisions, correlations, and connector boundaries.knowledgearchitecture:hermes-mac-mini-agent-osHermes Mac Mini Agent OSIntended local operator composition with explicit deployment-state, memory admission, browser, worker, interface, scheduler, and writeback proofs.knowledgeknowledge:capability-registryAgent Capability RegistryDiscovery inventory and promotion contracts for tools, skills, services, source artifacts, and external catalogs.knowledgeknowledge:agent-product-surfaceAgent Product SurfaceUser-facing intent, state, approvals, artifacts, proof, and public-agent sequencing across built-in and bring-your-own harnesses.knowledgeknowledge:build-dont-buyBuild, Don't BuyLayered decision rule for owning differentiated context and proof while renting replaceable infrastructure.knowledgecapability:cursor-sdkCursor SDKProgrammatic local and cloud Cursor agents used as a replaceable execution capability, not canonical brain ownership.capabilitycapability:hindsightHindsight Memory ProviderEvaluated runtime-memory candidate with retain, hybrid recall, explicit reflect, scoped banks, derived mental models, and fresh-target replay gates; Kevin-Wiki remains canonical.capabilitycapability:honchoHonchoPeer and profile modeling candidate kept separate from the single external runtime-memory slot and canonical wiki truth.capabilityinstruction:agent-operationsAgent Operations HubTask routing, read order, enforcement, and closeout.instructioninstruction:brain-operating-modelWiki Brain Operating ModelCanonical ownership and filing behavior.instructioninstruction:capability-routingCapability Routing MapDeterministic first-choice routes and fallback hierarchies for agent tools, skills, services, design, research, and execution.instructioninstruction:active-stackActive StackCurrent default, optional, guarded, and excluded tools and capabilities with explicit adoption boundaries.instructioninstruction:skill-resolverSkill ResolverDeterministic intent-to-skill and reference routing surface for agent tasks.instructionworkflow:source-compileSource CompileCapture, correlate, interrogate, write back, and prove source learning.workflowworkflow:automation-auditAutomation AuditInventory, classify, verify, reconcile, and repair automation ownership and freshness without treating a schedule declaration as proof of a running job.workflowworkflow:run-contractWorkflow Run ContractExecution, authority, proof, state, and writeback invariants.workflowworkflow:artifact-human-reviewArtifact Human ReviewExplicit router job for durable direct edits and anchored comments on local HTML, rendered Markdown, and localhost routes, with apply-before-ack proof and separate consequential authority.workflowworkflow:self-improving-pr-reviewSelf-Improving PR ReviewRead-only structured PR review plus a separate evidence-gated outer loop for reviewer improvement.workflowknowledge:loopyLoopyReusable goal, iteration, evaluator, and escalation-loop patterns for bounded agent improvement.knowledgeknowledge:agentic-quality-systemAgentic Quality SystemQuality ladders, critic separation, visual comparison, evidence, and stopping rules for high-ambition agent work.knowledgeknowledge:agent-self-improvement-evalsAgent Self-Improvement Eval LibraryDeterministic receipts, controlled evaluator epochs, bounded skill optimization, and paired-objective guardrail-policy evolution with frozen runs, unguarded baselines, utility floors, rollback, and reviewed promotion.knowledgeskill:agent-eval-libraryAgent Eval LibraryExecutable receipt, evaluator-evolution, skill-candidate, and guardrail-policy gates for agent self-improvement.skillworkflow:frontend-frontierFrontend Frontier RadarDesign and frontend evidence review routed into operating owners, including exact-source icon-system admission and rollback.workflowworkflow:career-pipelineCareer PipelineEvidence-backed role discovery, triage, evaluation, drafting, human decision, state transition, and proof-bearing writeback.workflowskill:absorb-sourcesAbsorb SourcesPortable agent procedure for source capture, interrogation, and proof.skillskill:hermes-operator-stackHermes Operator StackAudits and configures the persistent operator stack with deployment-state, memory admission, bank isolation, cost, leakage, recovery, and canonical-writeback proof.skillskill:skill-creatorSkill CreatorCreates and evolves executable skills through generated projections, bounded patches, protected slow policy, separate router/body evidence, strict held-out selection, lifecycle gates, and rollback.skillskill:frontend-design-tasteFrontend Design TasteExecutable art-direction and anti-generic frontend route with product-specific icon-system admission instead of blanket package swaps.skilldesign:systemDesign SystemTypography, icon systems, motion, structure, interaction, and proof rules.design-rulephilosophy:visible-systemsVisible Systems as Design LanguageKevin's personal design philosophy: visible construction, one owner per relationship, constrained expression, conserved motion, configurable instruments, and auditable taste.knowledgedesign:ai-native-patternsAI-Native Design PatternsProduction patterns for agent run truth, human authority, context and evidence, intent entry, and replayable stateful UI.design-ruledesign:astryxAstryx Design SystemExact-version route for Astryx as a Meta, React 19, agent-readable design-system reference and evidence-gated project candidate.design-ruleknowledge:ui-design-defaultsUI Design DefaultsConstruction defaults plus exact-source icon-family coverage, licensing, wrapper, accessibility, bundle, and rollback decisions.knowledgeknowledge:creative-web-interactionsCreative Web Interaction ReferencesEvidence-backed interaction vocabulary and behavioral contracts for expressive, input-driven, collaborative, generative, and community web surfaces.knowledgedesign:methodologyDesign Methodology AtlasRoutes discovery, product-model, system, and evaluation methods, including the seven-layer diagnostic from observed behavior through surface.design-ruleskill:design-engineering-polishDesign Engineering PolishExecutable motion and component-polish route for building, auditing, planning, opportunity discovery, prototyping, and library selection.skillskill:scroll-worldScroll WorldPinned executable procedure for generating a paid-provider, frame- and velocity-continuous, scroll-scrubbed cinematic world with native-mobile and reduced-motion proof.skillskill:illoilloPinned recurring-character editorial illustration procedure with identity references, visual QA, replay manifests, and fail-closed provider cost boundaries.skillcapability:illoilloReviewed tool owner for recurring-character editorial illustration, exact upstream provenance, compatibility, and adoption boundaries.capabilityskill:liteparseLiteParseExecutable parse-once document extraction route with bounded text search, structure-preserving Markdown, coordinate-bearing JSON, and visual escalation.skillskill:career-opsCareer OpsProgressive career-operation router for discovery, evaluation, drafting, pipeline state, interviews, and offer preparation with human authority gates.skillskill:social-contentSocial ContentExecutable social-content route from voice foundation and one canonical source through channel adaptation, approval, and evidence writeback.skillproject:fieldworkFieldworkInstallable design IDE and agent-connected visual inspection environment.projectproject:daily-briefDaily BriefInstalled local review and execution surface for ranked proposals, exact decisions, confined verification, and commit or PR delivery.projectcapability:governed-public-agentGoverned Public AgentPortable projection-bound agent contract that publishes read-only next steps and lets one scheduled, identity-scoped producer open a private Daily Brief proposal without mutation authority.capabilityproject:local-searchLocal Search CLIAgent-callable local-browser search with structured output and cleanup.projectproject:agent-broomAgent BroomStandalone cleanup and process-hygiene CLI for agents.projectcapability:deepsecDeepsecAgent-powered repository security analysis with benchmark-governed model routing, calibrated cost, refusal receipts, revalidation, and explicit approval gates.capabilitycapability:impeccableImpeccableReproducible frontend direction, design intent, live iteration, and deterministic review for greenfield, redesign, addition, and refinement work.capabilitycapability:tasteTasteConsolidated anti-slop and visual-direction capability with folded upstream references, review-gated updates, and one executable local owner.capabilitycapability:superpowersSuperpowersUpstream skills methodology whose design, planning, review, and proof invariants are locally absorbed without claiming the plugin is installed.capabilitycapability:transitionsTransitions.devReviewed local CSS transition recipes plus a guarded upstream update path that separates content-only patterns from Pro and Refine mutation surfaces.capabilitycapability:ui-skill-directoryUI SkillsExternal design-skill discovery shelf with selective routing, including UI UX Pro Max as a candidate design-intelligence source rather than house authority.capabilitycapability:pdf-operationsPDF OperationsPrivacy- and deployment-aware routing across PDF extraction, PDFCraft browser-local workflows, and Stirling-PDF self-hosted automation.capabilitycapability:liteparseLiteParseFast local model-free document parsing across text, Markdown, JSON, screenshots, and bounding boxes with benchmark and complexity boundaries.capabilitycapability:career-opsCareer-OpsPortable local-first career operating system with public-source discovery, fit evaluation, artifact drafting, state tracking, and explicit human authority.capabilityworkflow:security-scanSecurity ScanManual Deepsec scan route with source trust checks, live profile selection, cost calibration, verified findings, and model-selection proof.workflowskill:deepsecDeepsecExecutable procedure for selecting, calibrating, running, revalidating, and exporting a Deepsec security audit.skillskill:pdf-operationsPDF OperationsCallable procedure for selecting and safely executing PDF extraction, mutation, OCR, redaction, and repeatable automation with verification receipts.skillcapability:deterministic-videoDeterministic VideoRoutes reproducible agent-authored video through HyperFrames by default, with explicit Remotion, Lottie, and generative-video exceptions.capabilitycapability:vercel-eveVercel EveFilesystem-first durable agent framework with tools, skills, channels, connections, schedules, sandboxes, subagents, evals, and run observability.capabilityarchitecture:agent-company-osAgent Company OSOrganization-scale agent model for typed context planes, action-level authority, workflow eligibility, bounded execution, proof, recovery, and admitted learning.knowledgearchitecture:agent-control-planeAgent Control Plane ArchitectureFleet desired state, identity, routing, schedules, policy, approval, cost, observation, recovery, and replaceable execution workers.knowledgecapability:paperclipPaperclipPilot-gated self-hosted organizational control plane for persistent heterogeneous agents, goals, tasks, budgets, routines, approvals, and workspaces.capabilitycapability:code-review-graphCode Review GraphConditional repository-local Tree-sitter and MCP code graph used after the Graphify-first route when project-owned structural context is warranted.capabilityknowledge:agent-tool-source-ledgerAgent Tool Source LedgerCorrelated X, repository, media, and primary-source decisions for agent tools that reinforce, refine, or remain on guarded watch without page proliferation.knowledgecapability:agentosagentOSLightweight isolated agent execution substrate and reference implementation for separating durable harness state from replaceable workspaces.capabilityknowledge:frontierFrontier Signal InboxCurrent external evidence awaiting or reinforcing object interrogation.knowledgeknowledge:agent-sandboxesAgent SandboxesIsolation, state, developer workflow, and provider choices for agent execution environments.knowledgeknowledge:retrieval-qualityRetrieval QualityEvaluation contract for lexical, semantic, structural, visual, query-adapted, freshness-aware, and provenance-preserving retrieval.knowledgeknowledge:human-in-the-loop-controlHuman-in-the-Loop ControlHuman approval and typed control verbs backed by pre-model mechanical gates, sensitive-data minimization, fail-closed deferral, frozen inputs, resumable state, decision receipts, and realtime voice wait/cancel/expiry semantics.knowledgeknowledge:agent-memory-patternsAgent Memory PatternsLossless retention, selective promotion, memory types, late filtering, temporal correctness, abstention, and behavior-changing recall.knowledgearchitecture:agent-memory-systemAgent Memory System ArchitectureEvidence-to-context layers, storage lanes, write and read paths, conflict handling, module-level evaluation, and local maintenance.knowledgecapability:portable-memory-writesPortable Authenticated Memory WritesBearer-authenticated, authority-scoped remember, recall, and retire contract that stores provenance-bound owner proposals in a hash-chained private ledger without rewriting canonical wiki truth.capabilityknowledge:agent-native-memory-researchAgent-Native Memory System ResearchPaper-backed four-module memory framework, workload findings, component ablations, cost boundary, and Kevin Wiki implications.knowledgecapability:memorydataMemoryDataHeld research benchmark candidate for memory-system comparison with explicit licensing, dataset, endpoint, cost, and reproducibility gates.capabilityknowledge:eval-loopThe Eval LoopOutput-quality measurement, regression, runtime and production gates, failure feedback, scientific experiments, and checkpointed evaluator evolution.knowledgeworkflow:research-experimentResearch Experiment WorkflowFrozen-baseline research, data audit, exact-source pilot, monitored ablation, held-out confirmation, and provenance-bearing writeback.workflowcapability:ml-internML InternGuarded Hugging Face research agent for literature, datasets, implementation, sandbox and Jobs execution, evaluation, and ablation.capabilitycapability:website-reference-reconstructionWebsite Reference ReconstructionAuthorized live-site structural reconstruction through DOM, computed CSS, behavior extraction, isolated implementation, visual comparison, and independent rights/accessibility gates.capabilitycapability:recordlyRecordlyHuman-operated cross-platform screen capture and polished demo editing with an explicit third-party extension trust boundary.capabilitycapability:agent-reachAgent ReachLeast-privilege multi-platform research router over channel-specific public and authenticated tools.capabilitycapability:skillclawSkillClawGuarded session-capture and skill-proposal reference whose proxy, data, evaluation, promotion, versioning, and rollback boundaries block global installation.capabilitycapability:open-generative-aiOpen Generative AIOpen-source creative UI and local-inference reference with explicit MuAPI egress, cost, media-rights, and recipe-installation boundaries.capabilitycapability:firecrawlFirecrawlStructured web scrape, search, map, crawl, batch, monitoring, and interaction escalation with hosted-data, credit, self-host auth, and selective-install controls.capabilitycapability:kimi-k3Kimi K3Guarded multimodal, million-token, open-weight model candidate whose promotion requires same-harness receipts plus provider, privacy, license, cost, and hardware gates.capabilityknowledge:clipping-distributionClipping as Distribution StrategyRights-cleared source sessions compiled into curated, policy-safe, attributed short-form variants with explicit hypotheses, approvals, measurement, and graduation gates.knowledgeworkflow:content-pipelineContent Pipeline WorkflowEvidence-driven content planning and a bounded clipping branch from canonical source through curation, controlled variation, approval, publication, observation, and writeback.workflowknowledge:skill-evolutionControlled Skill Evolution - GEPA and SkillOptGEPA and SkillOpt synthesis with frozen experiments, bounded candidate patches, protected slow state, separate router/body/end-to-end evidence, strict held-out selection, rejected-edit memory, and reviewed rollbackable promotion.knowledgeworkflow:skill-syncSkill Sync WorkflowCanonical source-to-registry-to-runtime skill projection, including gated promotion for session-derived candidates.workflowcapability:research-workspacesSelf-Hosted Research WorkspacesRoutes bounded source-grounded notebook work to Open Notebook while keeping wiki and qmd canonical; reserves Odysseus for an isolated high-authority pilot.capabilitycapability:web-to-nativeWeb-to-Native ApplicationsRoutes native web shells from PWA through narrow project-owned Zero Native bridges, with WebToApp confined to an isolated Android test-device workshop.capabilitycapability:authorized-media-intakeAuthorized Media IntakeRoutes reproducible agent capture through direct yt-dlp and human batch intake through loopback-only ReClip with rights and provenance receipts.capabilityknowledge:project-catalogProject CatalogIdeas and projects that need deliberate scope and advancement decisions.knowledgecapability:browser-automationAgent Browser AutomationDefault browser exploration and verification route, with guarded escalation to reviewed reusable authenticated adapters.capabilityknowledge:component-sourcesComponent Library SourcesRanked component, primitive, registry, icon, chart, and effect sources with adoption, license, dependency, and accessibility boundaries.knowledgecapability:toolcraftToolcraftSource-owned creative-app runtime for agent-built canvas, control, animation, reference-study, export, and proof workflows; distinct from ordinary component-library routing.capabilitycapability:react-bitsReact BitsExpressive React component and effect source with preview, code, prompt, parameter, dependency, accessibility, and Commons Clause adoption boundaries.capabilitycapability:originkitOriginkitGuarded exact-effect source for hosted Framer/React components with separate preview, registry, MCP/CLI, delivered-payload, licensing, accessibility, lifecycle, and performance boundaries.capabilityknowledge:motion-effects-sourcesDesign Effects and Motion SourcesMotion, effect, shader, cursor, texture, and creative-interaction sources routed as engines, components, or bounded pattern studies.knowledgecapability:scroll-worldScroll WorldExact-source route for pre-rendered scroll-cinematic landing worlds, separated from runtime 3D, DOM choreography, and standalone video with provider, spend, seam, mobile, and proof boundaries.capabilitydesign:shader-effects-librariesShader Effects LibrariesCanonical shader, holographic-surface, material-workbench, and implementation-admission owner with physical-model, runtime, export, performance, and license boundaries.design-ruleknowledge:design-inspirationDesign Inspiration GalleriesCurrent and durable design discovery routes separated by craft, shipped-product flow, interaction, niche, and source-authority role.knowledgeskill:animated-component-librariesAnimated Component LibrariesExecutable source-selection procedure for primitives, registries, charts, effects, icons, and guarded component adoption before hand-rolling.skillknowledge:dev-tool-source-ledgerX Bookmark Developer Tool LedgerReplay ledger for developer-tool bookmarks, retained evidence, routing decisions, and graduation into canonical operating owners.knowledgeknowledge:ui-skills-source-ledgerUI Skills Source LedgerPractitioner rankings and live UI-skill sources correlated to canonical local routes without duplicate installs or pages.knowledgeskill:make-interfaces-feel-betterMake Interfaces Feel BetterExecutable interface-polish review covering typography, surfaces, motion, icons, performance, and complete state proof.skillskill:gstack-design-reviewGstack Design ReviewRendered surface audit and change-scoped interface review with affected-consumer expansion, removed-line regression detection, intent checks, provenance classification, and read-only PR inspection.skillskill:human-reviewHuman ReviewPinned local visual feedback loop for exact user edits and anchored comments, with durable batches, source mapping, stale-write protection, explicit acknowledgement, and separate approval authority.skillskill:autoreviewAuto ReviewIsolated structured code review across standards and spec axes with scope control, evidence-bound findings, and bounded improvement cycles.skillskill:agent-iteration-loopAgent Iteration LoopEnd-to-end implementation loop with separate cleanup and verification stages, live proof, review, CI triage, and controlled shipping.skillskill:prPull RequestRepository-native pull-request creation and refresh with exact proof, template resolution, and preservation of human context and media artifacts.skillskill:triageTriageEvidence-first issue and external-PR triage with fix-level clustering, credited consolidation, dry-run authority, and agent-ready briefs.skillskill:accessibilityAccessibilityWCAG 2.2 audit and targeted remediation route with native-semantics, keyboard, focus, contrast, media, and motion proof.skillskill:react-doctorReact DoctorChanged, full, design, and security scans with live triage, rule-level guidance, repository checks, and score receipts.skillskill:vercel-react-best-practicesVercel React Best PracticesCurrent prioritized React and Next.js performance rules retained under the Vercel source family.skillknowledge:actionable-tools-indexActionable Tools IndexIntent-first registry of retained tools with invocation, licensing, deployment, and operating boundaries.knowledgeknowledge:local-coding-agentsLocal Coding AgentsModel, runtime, harness, hardware, and same-task proof boundaries for local and open-weight coding agents.knowledgecapability:workflow-automationWorkflow Automation PlatformsRoutes canonical code-owned workflows, project-scoped n8n, and managed Gumloop by auditability, connector, authority, and operational needs.capabilitycapability:seo-geoSEO + GEO MeasurementRoutes search and AI visibility evidence, crawler controls, programmatic generation/index admission, first-party outcomes, failure recovery, and vendor claims without collapsing them into one rank.capabilityskill:ai-seoAI SEOExecutable AI-search audit and optimization procedure with current Search Console measurement, per-surface crawler policy, evidence boundaries, and repeated prompt/referral proof.skillskill:programmatic-seoProgrammatic SEOExecutable page-family route from demand and typed taxonomy through schema validation, page-type rendering, staged index admission, cohort proof, maintenance, and scaled-content recovery.skillworkflow:seo-geo-radarSEO GEO Radar WorkflowReplays search and AI-visibility claims against current first-party policy, site state, outcome evidence, and programmatic-page failure history before writeback.workflowknowledge:ai-seo-platform-ranking-factorsAI Search Platform Access and EvidenceFirst-party platform access routes, crawler distinctions, safe experiments, and anti-guarantee boundaries for Google, OpenAI, Perplexity, Microsoft, and Anthropic surfaces.knowledgeknowledge:seo-geo-evaluation-checklistSEO/GEO Evaluation ChecklistAudience-first technical, crawler, content, measurement, experiment, agent-usability, and proof gates without bot conflation, fixed-format folklore, or screenshot causality.knowledgecapability:coder-apps-routingCoder Apps RoutingJob-shaped routes for conversational forms, self-hosted availability monitoring, managed PaaS, standalone agent-readable docs, and editorial product discovery.capabilitycapability:coolifyCoolifyRecommended self-hosted PaaS route with owned-server operations, team-scoped MCP observation, least-privilege CLI/API mutation, and backup, restore, canary, health, rollback, and escape proof.capabilityknowledge:modern-web-vpsModern Web VPS StackRoute-first TanStack, Tailwind, Vite, Hetzner, Cloudflare, and optional Coolify stack with explicit ownership, agent authority, recovery, and production proof.knowledgeskill:service-cli-registryService CLI RegistryExecutable horizontal SaaS and infrastructure interface router with account/context, read-first, scoped-authority, preview, mutation, and receipt gates.skillcapability:glm-5-2GLM-5.2Guarded open-weight long-horizon model candidate with separate weight/code licenses and model-provider-effort-harness admission receipts.capabilityworkflow:cross-harness-model-routingCross-Harness Model RoutingRoutes judgment, execution, and verification by capability while admitting new model/provider/effort tuples only through frozen-condition comparative receipts.workflowcapability:hermes-agentHermes AgentPersistent self-owned agent runtime with skills, memory, automations, gateways, approvals, model routing, and proof-bearing upgrade boundaries.capabilityarchitecture:llm-wiki-harnessLLM Wiki Agent HarnessCross-runtime brain architecture joining content-addressed evidence, stable objects, named workflows, skills, authority, proof, retrieval, graph, API, and UI projections.knowledgecapability:pixelragPixelRAGConditional screenshot-tile retrieval lane for layout-bearing evidence, governed by render-state, region/link grounding, hybrid evaluation, privacy, rights, moderation, model, and cost receipts.capabilityknowledge:agent-search-and-retrievalAgent Search and RetrievalSource-of-truth search router spanning compiled wiki, code topology, current docs and web evidence, skills discovery, and receipt-bearing hybrid visual retrieval.knowledgecapability:camofox-browserCamofox BrowserGuarded Camoufox server route for authorized protected-page retrieval after less-evasive controls, failure classification, safe runtime configuration, and target-specific quality proof.capabilityknowledge:scraping-stealthAuthorized Protected-Page RetrievalLayered transport, browser, protocol, network, challenge, authority, security, and evaluation contract for permitted retrieval from defended websites.knowledgecapability:unbrokerUnbrokerGuarded personal data-broker removal reference: official DROP first, then assisted encrypted gap handling under scoped authority, field-level approval, mixed-license review, and external removal proof.capabilitycapability:hugging-face-speech-to-speechHugging Face Speech to SpeechConditional modular Python and OpenAI-Realtime-compatible voice pipeline with frozen VAD, STT, LLM, TTS, transport, release, model-license, security, privacy, and measured admission receipts.capabilityworkflow:realtime-voice-agentRealtime Voice Agent WorkflowRoutes realtime voice and wearable systems through model, sensor, data/tool, action-authority, and product-proof gates; separate perception/action loops; current session/model routes; human-only confirmation; durable work; latency, quality, safety, privacy, cost, accessibility, and rollback evidence.workflowcapability:visionclawVisionClawExact-revision wearable voice-and-vision architecture reference separating current iOS LiveKit perception and hosted action planes from the stale README/Android OpenClaw path, with license, confirmation, identity, persistence, and proof gates.capabilitycapability:vercel-ai-sdkVercel AI SDKTypeScript and React AI application toolkit for streaming, agents, durable workflows, harness adapters, AI Gateway routing, realtime voice, tests, and UI integration.capabilitycapability:moneyprinter-turboMoneyPrinterTurboRecommended high-throughput stock-footage narrated-short assembly route with frozen runtime, claim, media/music/voice rights, local security, provider/cost, complete-review, and explicit-publication receipts.capabilitycapability:hyperframesHyperFramesRecommended deterministic agent-authored video route for bespoke launch, demo, explainer, PR, social, code, data, and brand-motion scenes with source, render, and complete visual/audio proof.capabilityknowledge:ui-component-vocabularyUI Component VocabularyFuzzy visual description to platform-verified component or style name, confusable distinction, API/semantic symbol, and prompt-ready behavior.knowledgeskill:ui-vocabularyUI VocabularyProbationary callable route for naming unfamiliar UI from text, screenshots, rendered behavior, accessibility state, DOM, or code while preserving private-query boundaries and primary-source proof.skillskill:img2threejsimg2threejsPinned external procedure for reconstructing one suitable reference image as resumable, quality-gated, editable, animation-ready procedural Three.js code with explicit single-view uncertainty and visual proof.skillknowledge:web3d-routingWeb3D Library RankingRoutes reference-image procedural reconstruction, Three.js, R3F, engines, shaders, no-code WebGL, vector 3D, and asset pipelines by the actual 3D job and proof boundary.knowledgeknowledge:animations-devAnimations.dev And Emil Design-Engineering SkillsRevision-pinned source authority for Emil Kowalski's animation and design-engineering methods, with ten upstream branches folded into stable local owners and tested routes instead of duplicated global skills.knowledgecapability:santander-aiSantander AI Open SourceExact-revision router across Santander AI's 14 public repositories, with conditional adoption, governance-policy conflicts, observed release proof, and patterns integrated into Kevin's eval, HITL, retrieval, and source-compile owners.capabilitycapability:shopifyShopifyGoverned theme and store route separating Liquid source, official documentation and validation, Admin data, customer commerce, previews, and publication authority.capabilityskill:shopify-commerceShopify CommerceExecutable Shopify theme, Liquid, CLI, Admin, preview, accessibility, commerce-proof, rollback, and human-publication procedure.skillskill:wayfinderWayfinderManual-only multi-session decision discovery through a local-first map, decision tickets, acyclic blocking edges, a computed frontier, fog of war, and one decision per session.skillcapability:engineering-skill-sourcesEngineering Skill SourcesExact-revision router across Matt Pocock and David Ondrej engineering/orchestration procedures, preserving all signals while promoting only distinct, safe, testable local jobs.capabilitycapability:loreLore Version ControlEpic Games' MIT, Rust-led, binary-first centralized/offline-capable version-control candidate, routed only for measured large-asset workloads under a pinned pre-1.0 pilot and recovery gate.capabilitycapability:openauthOpenAuthRetained beta self-hosted OAuth provider for teams that explicitly own user, token, password, email, key, abuse, recovery, audit, migration, and production security operations.capabilitycapability:supabase-tanstack-dbSupabase TanStack DBExperimental typed optimistic/live Supabase collection adapter, retained for bounded capped prototypes behind RLS, conflict, reconciliation, SSR, volume, cost, and rollback proof.capabilitycapability:effect-tsEffect TypeScriptConditional typed-effect and structured-concurrency route for TypeScript systems with recurring lifecycle complexity; v3 is stable and v4 remains a separately evaluated beta.capabilitycapability:agentenvAgentENVConditional Linux/KVM Firecracker fleet with an E2B-compatible API, retained behind independent authentication, isolation, lifecycle, snapshot, recovery, and reproduced performance proof.capabilitycapability:open-code-reviewOpenCodeReviewOptional deterministic diff-selection, path-rule, and file-coverage ledger beneath Auto Review for large or heterogeneous changes.capabilitycapability:authenticated-task-spacesAuthenticated Browser Task SpacesGuarded ego-lite route for explicit agent/user ownership and handoff inside long-lived authenticated browser spaces; agent-browser remains default.capabilitycapability:visual-cmsSelf-Hosted Visual CMSConditional Instatic route and architecture reference for one editable page tree, agent/visual projections, capability gates, plugin isolation, and static output.capabilitycapability:cad-fabricationCAD, Robotics, and FabricationPinned local skill family for STEP-first CAD, drawings, robot descriptions, visual inspection, slicing, vendor preflight, and gated physical handoff.capabilityworkflow:cad-fabricationCAD and Fabrication WorkflowRoute digital geometry and robot-description work through explicit units, editable source, deterministic validation, visual proof, and separate physical authority.workflowskill:cadCADSTEP-first parametric part and assembly generation, inspection, validation, and secondary exports.skillskill:cad-viewerCAD ViewerLocal visual review links for explicit CAD, implicit CAD, drawings, and robot-description files.skillskill:step-partsSTEP PartsCanonical off-the-shelf CAD part discovery and checksum-verified STEP retrieval.skillskill:dxfDXFGenerated and validated two-dimensional manufacturing drawings from editable Python sources.skillskill:urdfURDFRobot-description authoring with explicit topology, frames, joints, limits, inertials, and geometry validation.skillskill:srdfSRDFMoveIt planning semantics, groups, end effectors, states, and disabled-collision validation over a valid URDF.skillskill:sdfSDFormatSDFormat model and world authoring, validation, and simulator handoff.skillskill:implicit-cadImplicit CADExperimental browser-native signed-distance geometry route used only when explicitly requested.skillskill:gcodeG-codePrinter-profiled slicing, dry-run inspection, and static validation without printer-specific execution.skillskill:sendcutsendSendCutSend PreflightConservative evidence-backed DXF and STEP upload, material, and service preflight reports.skillskill:bambu-labsBambu LabsDry-run-first local Bambu upload and explicit-confirmation print start from validated G-code.skillcapability:claude-dynamic-workflowsClaude Code Dynamic WorkflowsClaude-generated JavaScript orchestration and official Agent SDK fan-out/adversarial-verification fixtures beneath Kevin's cross-harness workflow registry.capabilitycapability:claude-managed-agentsClaude Managed AgentsHosted stateful-agent runtime and official applied cookbook patterns retained behind local identity, authority, data, budget, recovery, and proof contracts.capabilityprotocol:mcp-2026MCP 2026 Protocol EraFinal 2026-07-28 stateless MCP core with explicit application state, Apps and Tasks extensions, authorization hardening, version negotiation, and conformance-gated adoption.knowledgeskill:grill-with-docsGrill With DocsShape-first planning that resolves a dependency-aware decision frontier in rounds while codifying domain language, durable decisions, and prototype evidence.skillskill:handoffHandoffPortable context transfer for a changed recipient, harness, directory, project, or human, with canonical artifacts referenced rather than duplicated.skillknowledge:design-treeDesign-Tree ExplorationShape the decision graph before depth, then resolve all currently unblocked decisions through dependency-frontier rounds.knowledgeworkflow:ai-development-phasesAI-Powered Development PhasesSeven-phase alignment, research, prototype, specification, ticket, implementation, and review workflow with explicit phase exit gates.workflowknowledge:font-picksFree Font PicksLicensed, categorized font shelf with exact family sources, high-character libre serifs, fit guidance, and per-project typography proof.knowledgearchitecture:daily-brief-control-planeDaily Brief Control PlaneAppend-only event, proposal, decision, source-cursor, delivery-ledger, privacy, replay, and adapter contract beneath the Daily Brief product.knowledge