Concept System Map

Concepts are reusable mental models: they explain a primitive that transfers across projects, tools, workflows, skills, architectures, or decisions.

Use this page to choose which concept article to read or update. Use wiki/concepts/README.md for the local directory router and Concepts Filing for the filing rule. A concept page should make future action clearer; if it only names a topic, it belongs as a section, table row, generated view, or redirect.

Concept Families

Family Canonical pages Owns
Brain, memory, and retrieval Knowledge Compilation, The Brain-Agent Loop, Agent Memory Patterns, Retrieval-Augmented Agents, Retrieval Quality, Agent Search and Retrieval, Wiki-Net How knowledge becomes durable, searchable, linked, and useful to agents.
Agent harness and runtime primitives Agent Harness, Harness Engineering, Agent Looping, Browser Testing Skills, Agent Trajectory Evaluation, Agent Legibility, Agent Harness, Runtime, Memory, and Evals, Agent Harness Capsules What makes a model into an operator: loop, context, state, tools, memory, proof, and human visibility.
Capability, tool, and skill routing Agent Capability Registry, Skills, Tools, and Capability Discovery, Agent Skill Bundles, MCP as the Integration Standard, Agent-Native CLIs, Source-Code Registry Pattern How agents discover, route, trust, and progressively disclose executable capability.
Quality, proof, and self-improvement Proof-Carrying Work, Agentic Quality System, Doctor Pattern, The Eval Loop (Slop Is an Output Problem), Eval-Driven Development, Empirical Verification, Self-Maintaining Systems, Red Queen Gödel Machine, BINEVAL How work proves itself, improves over time, and avoids fluent but unverified output.
Security, safety, and production boundaries Agent Security Model, Production Safety, No-Fallbacks Policy, Containers Are Not a Security Boundary, npm Supply Chain Security, Lint-Enforced Agent Guardrails, Third-Party Noise Suppression How systems constrain authority, fail loudly, preserve evidence, and avoid unsafe side effects.
Infrastructure, sandboxes, and remote workers Agentic Infrastructure, Agent Sandboxes, Persistent Sandbox Thesis, Agent + Machine as the Primitive, Agents as Service Workers (as a Service), Tailnet Agent Mesh - Reaching Your Agents from Anywhere, Sandbox Startup Latency (Started ≠ Ready), Browser Agent Console Where agents run, how they persist, how they connect, and what "ready" means.
Browser, computer use, and rendered verification Browser Automation and QA, Computer Use and Browser Automation Patterns, Complementary Browser Testing, Local Web Verification, Browser Rendering Pipeline, Browser Testing Skills, Browser Agent Console When to use deterministic tests, live browser inspection, vision/browser agents, or content extraction.
Design engineering and interface quality Design Engineering, Constraints over Context, Inheritable Taste Through Skill Files, Taste Is the Differentiator (on Shared Primitives), UI Component Vocabulary, Sigil Visual Doctors, State-Aware Product Cards, Canvas Text Layout (Pretext), Performance Budgets How taste, UI systems, tokens, motion, performance, accessibility, and verification become executable.
Product, business, distribution, and positioning Services-as-Software, Open Core, Devtool Growth Playbook, Opinionated Templates, Insertion Layers, Algorithmic Slop Throttling (Posting for the X Algorithm), Clipping as Distribution Strategy, YC Verbs Positioning, Red-Team Your Business (Adversarial AI Audit) How products are framed, distributed, monetized, attacked, and differentiated.
Finance, tax, equity, and fundraising catalogs Finance Hacks, Tax Hacks Catalog, Individual Tax Hacks, Company Tax Hacks, Personal Finance Hacks, Startup Equity Hacks, Investment Vehicles, Pre-IPO & IPO Access Hacks, YC Standard Deal vs Negotiated Round Reference catalogs for money, taxes, equity, investment, and fundraising. These are not personalized advice.
Reasoning, prompting, and research abstractions Abstract Chain-of-Thought, Orchestration Variables, Structured Outputs Pipeline, Tool-Calling Design, Claude Prompting Principles, Adversarial Self-Audit Prompts, LLM Council, LLM Synthetic Panels (Purchase Intent) How models reason, expose structure, call tools, judge plans, and coordinate deliberation.
Generated skill-family synthesis Agent Engineering Skills, Agent Operations Skills, Frontend and Design Skills, Browser Testing Skills, Cloud, Data, and Service Skills, Firebase Skills, Azure Skills, Google Workspace Skills, Security and Review Skills, Writing and Content Skills, Media Generation Skills Generated family pages summarizing executable skill source. Source of truth remains skills/**/SKILL.md plus Skill Registry.

Merge Decisions

No existing concept pages were merged in the 2026-07-01 pass. The current near-duplicates are mostly layered concepts, not duplicate pages.

Cluster Decision Reason
Loop pages Keep distinct Agent Looping is the product/UX shift, Browser Testing Skills is an implementation workflow, The Eval Loop (Slop Is an Output Problem) is a quality loop, and The Brain-Agent Loop is the wiki memory loop.
Harness pages Keep distinct, link through this map and Agent Harness Agent Harness, Harness Engineering, Agent Harness Capsules, and Agent Harness, Runtime, Memory, and Evals answer concept, discipline, capsule, and layer-synthesis questions.
Browser pages Keep distinct Browser QA, computer use, complementary testing, local verification, rendering pipeline, and browser-testing skills route different task shapes.
Skill-family pages Keep generated The *-skills concept pages are generated synthesis, not hand-written duplicates of executable skill source.
Finance pages Keep as hub plus catalogs Finance Hacks and Tax Hacks Catalog are routers; individual/company/equity/investment pages are specialized reference ledgers.
Architecture-like concepts Keep in concepts only when reusable Pages such as Persistent Sandbox Thesis and Agent + Machine as the Primitive are conceptual primitives; concrete system boundaries stay in architecture/.

Concept Page Contract

A concept article should answer five questions:

  1. What reusable primitive does this page own?
  2. What failure mode does the primitive prevent?
  3. Which projects, tools, workflows, skills, or architectures operationalize it?
  4. What adjacent pages should not be confused with it?
  5. What proof, source, or timeline evidence keeps the concept grounded?

If a page cannot answer those questions, fold it into a stronger concept, a directory router, a generated skill-family page, or a project/tool/architecture page.

Update Rule

When updating any concept page:

  • update its updated: date
  • keep a one-line lead
  • link it to the relevant concept family
  • preserve or add timeline evidence
  • connect the idea to operational pages, not only adjacent ideas
  • regenerate _index.md, backlinks, qmd, and generated packs when routing changes
  • log structural category work in wiki/log.md

Timeline

  • 2026-07-01 | Created as the category-level concept graph for the Concepts rewrite. Added concept families, no-merge decisions, and the page contract used to update all concept articles. Source: User request, 2026-07-01

98 pages link here

Adversarial Self-Audit PromptsConceptsAG-UI ProtocolConceptsAgent + Machine as the PrimitiveConceptsAgent Activity LogConceptsAgent Capability RegistryConceptsAgent Engineering SkillsConceptsAgent EthosConceptsAgent Harness, Runtime, Memory, and EvalsConceptsAgent Iteration LoopConceptsAgent Operations SkillsConceptsAgent Self-Improvement Eval LibraryConceptsAgent2Agent (A2A)ConceptsAgents as Service Workers (as a Service)ConceptsAlgorithmic Slop Throttling (Posting for the X Algorithm)ConceptsAutomation OrchestrationConceptsAzure SkillsConceptsBeta DX Walk (Friction Is a Bug)ConceptsBilling LedgersConceptsBINEVALConceptsBrowser Agent ConsoleConceptsBrowser Automation and QAConceptsBrowser Testing SkillsConceptsCanvas Text Layout (Pretext)ConceptsClipping as Distribution StrategyConceptsCloud, Data, and Service SkillsConceptsCompany Tax HacksConceptsComplementary Browser TestingConceptsConcepts FilingFiling Decision TreeConstraints over ContextConceptsContainers Are Not a Security BoundaryConceptsCredit Card & Points HacksConceptsCutover, Not CompatibilityConceptsDepth-of-Context InterfacesConceptsDesign-Tree Exploration (Shape Before Depth)ConceptsDoctor PatternConceptsDual-Audience ErrorsConceptsEmpirical VerificationConceptsEval-Driven DevelopmentConceptsExplicit I/O RuntimeConceptsFiling Decision TreeFiling Decision TreeFinance HacksConceptsFirebase and Azure Skill FamiliesConceptsFirebase SkillsConceptsFrontend and Design SkillsConceptsFrontend, Design, Motion, and AccessibilityConceptsGit WorkflowConceptsGoogle Workspace SkillsConceptsHarness-Beats-Model HypothesisConceptsHuman-in-the-Loop ControlConceptsIdentity, Session, and RoutingConceptsiMessage Wiki AccessConceptsIndividual Tax HacksConceptsInvariant-First TestingConceptsInvestment VehiclesConceptsKioto Bounded ConcurrencyConceptsKnowledge CompilationConceptsLatency Budgets for AgentsConceptsLive Process AuditingConceptsLive Runbook CanvasConceptsLocal Web VerificationConceptsMedia Generation SkillsConceptsMonitorability TaxConceptsNo-Fallbacks PolicyConceptsObservation Precedes OrchestrationConceptsOpen CoreConceptsOpinionated Templates, Insertion LayersConceptsOrchestration VariablesConceptsPersonal Finance HacksConceptsPlan Mode Review ChecklistConceptsPR Review Lessons (Dedalus)ConceptsPre-IPO & IPO Access HacksConceptsProduction SafetyConceptsPrompt-Caching EconomicsConceptsProof-Carrying WorkConceptsRecursive Agent OrchestrationConceptsRed Queen Gödel MachineConceptsRed-Team Your Business (Adversarial AI Audit)ConceptsSandbox Startup Latency (Started ≠ Ready)ConceptsSecurity and Review SkillsConceptsSelf-Maintaining SystemsConceptsSigil Visual DoctorsConceptsSkills, Tools, and Capability DiscoveryConceptsSource-Code Registry PatternConceptsSpec-Driven DevelopmentConceptsStartup Equity HacksConceptsState ResumabilityConceptsStaying in the Loop with AgentsConceptsStructured Outputs PipelineConceptsTax Hacks CatalogConceptsThe Brain-Agent LoopConceptsThird-Party Noise SuppressionConceptsTool-Calling DesignConceptsUser Model MapUSERWeb Share APIConceptsWiki Concept Coverage MapMetaWiki-NetConceptsWriting and Content SkillsConceptsYC Standard Deal vs Negotiated RoundConcepts