Brain Capsules

Public interface for exposing Kevin's compiled wiki brain to agents and user-facing workflows. Capsules own purpose, source pages, skills, tools, generated packs, and maintenance rules; packs are task bundles attached to capsules.

Generated: 2026-06-30 Manifest hash: 8c48ca05b60d Machine manifest: agent-docs/capsules.json

Contract

  • Capsules are the boundary agents should name when they use Kevin's brain.
  • Source wiki pages remain authoritative.
  • Generated packs are execution bundles attached to one or more capsules.
  • New bookmark insights, skills, tools, playbooks, diagrams, and version snapshots should update the relevant capsule source pages before packs regenerate.
  • Do not edit this generated page by hand; edit scripts/generate-agent-packs.ts, then run npm run agent-packs.

Capsule Table

Capsule Public purpose Read-first sources Skill routes Packs
identity-profile Expose Kevin's durable identity, preferences, cadence, and communication defaults without requiring the whole wiki. Agent Soul, User Profile, Operational Heartbeat, Identity, Session, and Routing, Agent Operations Hub kevin-voice, humanizer, agent-docs identity.md, session.md, routing.md, workflow-index.md, product.md, content.md, writing.md, seo.md, career.md, portfolio.md, people.md, principles.md, narrative.md
agent-operating-system Expose how Kevin's agents route work through skills, tools, memory, evals, automations, and proof. Agent Harness, Runtime, Memory, and Evals, Agent Harness Capsules, Agent Self-Improvement Eval Library, Red Queen Gödel Machine, Superpowers, Matt Pocock's Skills (Skills for Real Engineers) agent-docs, skill-creator, agent-iteration-loop, agent-eval-library, loopy identity.md, session.md, routing.md, workflow-index.md, architecture.md, ai-agents.md, skills.md, mcp-tools.md, memory.md, evals.md, sandboxes.md, model-routing.md, automations.md, wiki-ops.md, learning.md
claude-code-agent-stack Expose Kevin's compiled Claude Code knowledge: terminal harness behavior, hooks, skills, workflows, memory, rules, security, and cross-harness transfer patterns. Agent Harness Capsules, Claude Code Harness, Caveman Token Skill, Ponytail, Everything Claude Code - Agent Harness Performance System, Claude Code Dynamic Workflows agent-docs, skill-creator, claude-skill-creator, claude-find-skills, autoreview agent-harnesses.md
hermes-agent-stack Expose Kevin's compiled Hermes knowledge: persistent self-hosted agents, shareable profile distributions, skills as procedural memory, cron, approvals, goals, MoA, eval loops, and Agent Machines deployment. Agent Harness Capsules, Hermes Harness (Nous Research), Hermes Agent (Nous Research), Agent Machines, Hyperagent, Higgsfield Supercomputer agent-eval-library, loopy, skill-creator, agent-iteration-loop, x-bookmark-absorb ai-agents.md, agent-harnesses.md
openclaw-agent-stack Expose Kevin's compiled OpenClaw knowledge: gateway daemon, messaging channels, Nodes, AgentSkills, ClawHub, MCP/ACP, sandboxing, and skill-supply-chain risk. Agent Harness Capsules, OpenClaw Harness, Agent Machines, Cross-Agent Harness Portability, Agent Guardrails Architecture, MCP Server Architecture skill-auditor, agent-docs, security-threat-model, security-best-practices, browser-use agent-harnesses.md
codex-agent-stack Expose Kevin's compiled Codex knowledge: OpenAI coding harness, GPT-5.x-Codex, sandbox policy, AGENTS.md, MCP/App Server, automations, cloud threads, and Record & Replay skills. Agent Harness Capsules, Codex Harness (OpenAI), T3 Code, GPT-5-Codex, Codex Record & Replay, Security and Review Skills codex-automation-admin, agent-eval-library, autoreview, x-bookmark-absorb, agent-docs ai-agents.md, agent-harnesses.md
wiki-ingest-memory Expose the source-to-synthesis loop: raw capture, enriched artifacts, compiled pages, indexes, and durable memory. Wiki Brain Operating Model, Karpathy LLM Wiki Setup Guide, QMD - Local Wiki Search Engine, Capture Ingest Protocol, Cloud, Data, and Service Skills, Wiki Capsules And Agent Doc Packs x-bookmark-absorb, liteparse, wiki-doctor, agent-docs workflow-index.md, ai-agents.md, skills.md, mcp-tools.md, memory.md, evals.md, sandboxes.md, model-routing.md, automations.md, wiki-ops.md, capture-ingest.md, research.md, learning.md, docs-writing.md, diagrams.md
engineering-quality Expose Kevin's code taste, typed boundaries, review gates, doctors, security scans, and proof standards. Code Taste, Ponytail, Agent Engineering Skills, Zero Tech Debt, shadcn/improve, Borrowed Intelligence (Drain It into Plans) kevin-engineering-flow, zero-tech-debt, no-sus-code-doctor, autoreview, deepsec code.md, architecture.md, typescript.md, testing.md, debugging.md, review-qa.md, doctors.md, security.md, git-process.md, process-hygiene.md, react.md, browser-qa.md
frontend-design-system Expose Kevin's frontend, React, Next.js, StyleX, tokens, component-source, and browser verification defaults. Frontend, Design, Motion, and Accessibility, AI-Native Design Patterns, JSON Render -- Generative UI Framework, Copilot Design System, StyleX, v0 by Vercel frontend-design, web-design-engineer, animated-component-libraries, accessibility, playwright frontend.md, react.md, nextjs.md, design.md, design-systems.md, motion.md, accessibility.md, performance.md, browser-qa.md, web3d.md, mobile.md, product.md
taste-motion-polish Expose Emil Kowalski, inheritable taste, motion vocabulary, animation rules, and micro-polish sources. Emil Kowalski, Design and Animation, Inheritable Taste Through Skill Files, design-engineering-polish, Details That Make Interfaces Feel Better, UI Skills Pack design-engineering-polish, taste-skills-playbook, motion-framer, gsap-scrolltrigger, make-interfaces-feel-better code.md, review-qa.md, frontend.md, react.md, nextjs.md, design.md, design-systems.md, motion.md, accessibility.md, performance.md, browser-qa.md, web3d.md, mobile.md, skills.md, diagrams.md, media-generation.md, documents-presentations.md, game-dev.md, hardware-graphics.md, events-travel.md
security-and-supply-chain Expose Kevin's security routing, DeepSec calibration, dependency gates, threat models, and safe external tool adoption. Security and Review Skills, Security and Review Skills, Security and Review Skills, Security Scan Workflow, Security and Review Skills, Vercel Connect deepsec, no-sus-code-doctor, security-best-practices, security-threat-model, bugs code.md, architecture.md, typescript.md, testing.md, debugging.md, review-qa.md, doctors.md, security.md, git-process.md, process-hygiene.md, backend.md, databases.md, auth.md, infra.md, cloud.md, observability.md, payments.md, data.md, ops.md, service-cli.md, google-workspace.md
diagrams-visual-explanation Expose diagramming tools, Mermaid/Excalidraw routes, architecture maps, and visual explanation conventions. Hand-Drawn Diagrams, Security and Review Skills, Browser Testing Skills, Technical Illustration System, Native Mermaid Charts excalidraw-diagram-generator, diagram-to-html, azure-resource-visualizer architecture.md, wiki-ops.md, capture-ingest.md, research.md, learning.md, docs-writing.md, diagrams.md, portfolio.md, media-generation.md, documents-presentations.md, game-dev.md, hardware-graphics.md, events-travel.md
product-growth-playbooks Expose launch, content, SEO, product, devtool growth, portfolio, and reusable playbook knowledge. Devtool Growth Playbook, Launch Post Playbook, Frontend and Design Skills, Customer Feedback Emails, Product Site Reference Gallery, Application Standout Protocol content-strategy, social-draft, to-prd, to-issues, career-ops product.md, content.md, writing.md, seo.md, career.md, portfolio.md, people.md, principles.md, narrative.md, documents-presentations.md
backend-cloud-ops Expose service, cloud, data, auth, observability, payment, and safe external-operation routes. Cloud, Data, and Service Skills, Security and Review Skills, Production Safety, Google Workspace Skills, Observability Architecture, Supabase TanStack DB service-cli-registry, supabase, tanstack, firebase, azure-prepare nextjs.md, backend.md, databases.md, auth.md, infra.md, cloud.md, observability.md, payments.md, data.md, ops.md, service-cli.md, google-workspace.md

Identity And User Model

ID: identity-profile

Expose Kevin's durable identity, preferences, cadence, and communication defaults without requiring the whole wiki.

Load when: Any serious task, session start, handoff, or user-facing answer.

Read first

Skill routes

  • kevin-voice
  • humanizer
  • agent-docs

Tool routes

  • qmd
  • check-freshness

Generated packs

  • identity.md
  • session.md
  • routing.md
  • workflow-index.md
  • product.md
  • content.md
  • writing.md
  • seo.md
  • career.md
  • portfolio.md
  • people.md
  • principles.md
  • narrative.md

Maintenance

  • Update identity pages when durable preferences change.
  • Regenerate packs after identity/routing edits.

Agent Operating System

ID: agent-operating-system

Expose how Kevin's agents route work through skills, tools, memory, evals, automations, and proof.

Load when: Building, operating, or debugging agents and cross-harness workflows before choosing a runtime-specific capsule.

Read first

Skill routes

  • agent-docs
  • skill-creator
  • agent-iteration-loop
  • agent-eval-library
  • loopy
  • loop-me
  • codex-automation-admin

Tool routes

  • tool_search
  • qmd
  • skills.sh
  • Superpowers
  • vercel:eve
  • vercel:ai-sdk

Generated packs

  • identity.md
  • session.md
  • routing.md
  • workflow-index.md
  • architecture.md
  • ai-agents.md
  • skills.md
  • mcp-tools.md
  • memory.md
  • evals.md
  • sandboxes.md
  • model-routing.md
  • automations.md
  • wiki-ops.md
  • learning.md

Maintenance

  • Run agent-packs:audit after large wiki changes.
  • Promote repeated agent workflows into skills or automations.

Claude Code Agent Stack

ID: claude-code-agent-stack

Expose Kevin's compiled Claude Code knowledge: terminal harness behavior, hooks, skills, workflows, memory, rules, security, and cross-harness transfer patterns.

Load when: Using Claude Code, designing Claude skills/workflows/hooks/subagents, importing Claude patterns into another harness, or comparing terminal-first coding agents.

Read first

Skill routes

  • agent-docs
  • skill-creator
  • claude-skill-creator
  • claude-find-skills
  • autoreview
  • cleanup-terminals-browsers

Tool routes

  • Claude Code
  • claude
  • skills.sh
  • Superpowers
  • AgentShield
  • qmd

Generated packs

  • agent-harnesses.md

Maintenance

  • Promote Claude Code bookmarks into the harness/tool pages before regenerating packs.
  • Keep Claude-specific skills linked through the shared skill taxonomy instead of copying them into bespoke runtime folders.
  • Record current Claude Code version, workflow, hook, and managed-agent behavior when source docs change.

Hermes Agent Stack

ID: hermes-agent-stack

Expose Kevin's compiled Hermes knowledge: persistent self-hosted agents, shareable profile distributions, skills as procedural memory, cron, approvals, goals, MoA, eval loops, and Agent Machines deployment.

Load when: Building or operating persistent personal agents, self-hosted Hermes, profile distributions, eval-loop automations, cron monitors, /learn skill creation, or Hermes on Agent Machines.

Read first

Skill routes

  • agent-eval-library
  • loopy
  • skill-creator
  • agent-iteration-loop
  • x-bookmark-absorb
  • agent-docs

Tool routes

  • Hermes Agent
  • hermes
  • Higgsfield Supercomputer
  • MCP
  • ACP
  • Dedalus Machines
  • OpenRouter

Generated packs

  • ai-agents.md
  • agent-harnesses.md

Maintenance

  • Promote Hermes launch/docs/bookmark updates into hermes-agent and hermes-harness before pack regeneration.
  • Keep profile distributions versioned as public brain/config/skills while memories, sessions, and secrets stay outside the package.
  • Keep Higgsfield-style creative-agent product signals as architecture references, not default local dependencies.
  • Keep Agent Machines deployment notes synchronized with Hermes runtime assumptions.
  • Treat Hermes automation/eval claims as hypotheses until an eval-loop receipt exists.

OpenClaw Agent Stack

ID: openclaw-agent-stack

Expose Kevin's compiled OpenClaw knowledge: gateway daemon, messaging channels, Nodes, AgentSkills, ClawHub, MCP/ACP, sandboxing, and skill-supply-chain risk.

Load when: Working on OpenClaw, multi-channel assistants, ClawHub skills, OpenClaw security, message gateway architecture, or OpenClaw as an Agent Machines runtime.

Read first

Skill routes

  • skill-auditor
  • agent-docs
  • security-threat-model
  • security-best-practices
  • browser-use

Tool routes

  • OpenClaw
  • ClawHub
  • ~/.agents/skills
  • MCP
  • ACP
  • Codex supervisor

Generated packs

  • agent-harnesses.md

Maintenance

  • Treat third-party OpenClaw skills as untrusted supply-chain input.
  • Keep OpenClaw's skill bridge pointed at ~/.agents/skills rather than copying wiki skills.
  • Refresh OpenClaw docs/star/security facts before recommending it as a default runtime.

Codex Agent Stack

ID: codex-agent-stack

Expose Kevin's compiled Codex knowledge: OpenAI coding harness, GPT-5.x-Codex, sandbox policy, AGENTS.md, MCP/App Server, automations, cloud threads, and Record & Replay skills.

Load when: Using Codex CLI/app/cloud, configuring Codex automations, debugging Codex harness behavior, turning GUI work into skills, or comparing OpenAI's coding-agent runtime.

Read first

Skill routes

  • codex-automation-admin
  • agent-eval-library
  • autoreview
  • x-bookmark-absorb
  • agent-docs
  • cleanup-terminals-browsers

Tool routes

  • Codex
  • codex exec
  • Codex desktop
  • OpenAI Responses API
  • MCP
  • qmd

Generated packs

  • ai-agents.md
  • agent-harnesses.md

Maintenance

  • Run config-sync/bootstrap checks after Codex preset or skill-link changes.
  • Record Codex app/CLI/cloud feature versions for promoted changes.
  • Keep automations in sqlite as authority and TOML mirrors as secondary.

Wiki Ingest And Memory

ID: wiki-ingest-memory

Expose the source-to-synthesis loop: raw capture, enriched artifacts, compiled pages, indexes, and durable memory.

Load when: Adding links, bookmarks, raw drops, screenshots, articles, or recurring knowledge work.

Read first

Skill routes

  • x-bookmark-absorb
  • liteparse
  • wiki-doctor
  • agent-docs

Tool routes

  • qmd
  • build-index
  • audit:x-bookmark-artifacts
  • audit:x-bookmarks
  • lit

Generated packs

  • workflow-index.md
  • ai-agents.md
  • skills.md
  • mcp-tools.md
  • memory.md
  • evals.md
  • sandboxes.md
  • model-routing.md
  • automations.md
  • wiki-ops.md
  • capture-ingest.md
  • research.md
  • learning.md
  • docs-writing.md
  • diagrams.md

Maintenance

  • Run artifact audit and doctor --only x-bookmarks after X syncs.
  • Write back new concepts, tools, skills, playbooks, diagrams, and capsule updates.

Engineering Quality

ID: engineering-quality

Expose Kevin's code taste, typed boundaries, review gates, doctors, security scans, and proof standards.

Load when: Writing, reviewing, testing, debugging, securing, or shipping code.

Read first

Skill routes

  • kevin-engineering-flow
  • zero-tech-debt
  • no-sus-code-doctor
  • autoreview
  • deepsec
  • tests
  • tdd

Tool routes

  • shadcn/improve
  • react-doctor
  • deepsec
  • project-doctor
  • routing-doctor
  • gh

Generated packs

  • code.md
  • architecture.md
  • typescript.md
  • testing.md
  • debugging.md
  • review-qa.md
  • doctors.md
  • security.md
  • git-process.md
  • process-hygiene.md
  • react.md
  • browser-qa.md

Maintenance

  • Capture repeated code-review lessons as skills or rules.
  • Record version/security snapshots for promoted tools.

Frontend And Design System

ID: frontend-design-system

Expose Kevin's frontend, React, Next.js, StyleX, tokens, component-source, and browser verification defaults.

Load when: Building UI, design systems, React/Next surfaces, CSS architecture, or visual QA.

Read first

Skill routes

  • frontend-design
  • web-design-engineer
  • animated-component-libraries
  • accessibility
  • playwright
  • remocn

Tool routes

  • StyleX
  • Next.js
  • React
  • Playwright
  • Radix
  • shadcn
  • UI Skills
  • Elements
  • Efferd Dashboard Blocks
  • Astryx
  • Kumo
  • Motion Primitives
  • React Bits
  • Gradient Shimmer
  • Remocn
  • Realtime Colors
  • Haikei
  • Design Arena

Generated packs

  • frontend.md
  • react.md
  • nextjs.md
  • design.md
  • design-systems.md
  • motion.md
  • accessibility.md
  • performance.md
  • browser-qa.md
  • web3d.md
  • mobile.md
  • product.md

Maintenance

  • Run browser screenshots for UI changes.
  • Promote agent-era design-system constraints into lint/type gates.

Taste, Motion, And Polish

ID: taste-motion-polish

Expose Emil Kowalski, inheritable taste, motion vocabulary, animation rules, and micro-polish sources.

Load when: Making UI feel better, choosing motion, reviewing animation, or creating visual style.

Read first

Skill routes

  • design-engineering-polish
  • taste-skills-playbook
  • motion-framer
  • gsap-scrolltrigger
  • make-interfaces-feel-better
  • remocn
  • remotion-best-practices

Tool routes

  • animations.dev
  • UI Skills
  • Motion
  • Motion Primitives
  • Gradient Shimmer
  • Remocn
  • GSAP
  • Anime.js
  • Haikei
  • Realtime Colors

Generated packs

  • code.md
  • review-qa.md
  • frontend.md
  • react.md
  • nextjs.md
  • design.md
  • design-systems.md
  • motion.md
  • accessibility.md
  • performance.md
  • browser-qa.md
  • web3d.md
  • mobile.md
  • skills.md
  • diagrams.md
  • media-generation.md
  • documents-presentations.md
  • game-dev.md
  • hardware-graphics.md
  • events-travel.md

Maintenance

  • Add approved visual decisions to taste skills.
  • Keep motion guidance tied to reduced-motion and performance rules.

Security And Supply Chain

ID: security-and-supply-chain

Expose Kevin's security routing, DeepSec calibration, dependency gates, threat models, and safe external tool adoption.

Load when: Auth, secrets, webhooks, dependency installs, repo scans, MCP/tool adoption, or production release risk.

Read first

Skill routes

  • deepsec
  • no-sus-code-doctor
  • security-best-practices
  • security-threat-model
  • bugs
  • brin

Tool routes

  • deepsec
  • React Security Doctor
  • SkillSpector
  • no-sus-code-doctor
  • brin
  • npm audit
  • CodeQL
  • Semgrep

Generated packs

  • code.md
  • architecture.md
  • typescript.md
  • testing.md
  • debugging.md
  • review-qa.md
  • doctors.md
  • security.md
  • git-process.md
  • process-hygiene.md
  • backend.md
  • databases.md
  • auth.md
  • infra.md
  • cloud.md
  • observability.md
  • payments.md
  • data.md
  • ops.md
  • service-cli.md
  • google-workspace.md

Maintenance

  • Calibrate cost before full DeepSec scans.
  • Record security scan proof and accepted residual risk.

Diagrams And Visual Explanation

ID: diagrams-visual-explanation

Expose diagramming tools, Mermaid/Excalidraw routes, architecture maps, and visual explanation conventions.

Load when: Creating diagrams, architecture maps, flowcharts, visual docs, or artifact explanations.

Read first

Skill routes

  • excalidraw-diagram-generator
  • diagram-to-html
  • azure-resource-visualizer

Tool routes

  • Mermaid
  • Excalidraw
  • diagram-to-html

Generated packs

  • architecture.md
  • wiki-ops.md
  • capture-ingest.md
  • research.md
  • learning.md
  • docs-writing.md
  • diagrams.md
  • portfolio.md
  • media-generation.md
  • documents-presentations.md
  • game-dev.md
  • hardware-graphics.md
  • events-travel.md

Maintenance

  • Render or inspect diagrams before handoff.
  • Keep diagram source files linked from the page that explains them.

Product, Growth, And Playbooks

ID: product-growth-playbooks

Expose launch, content, SEO, product, devtool growth, portfolio, and reusable playbook knowledge.

Load when: Planning products, launches, content, career artifacts, growth motions, or strategy docs.

Read first

Skill routes

  • content-strategy
  • social-draft
  • to-prd
  • to-issues
  • career-ops

Tool routes

  • qmd
  • Google Workspace
  • LinkedIn

Generated packs

  • product.md
  • content.md
  • writing.md
  • seo.md
  • career.md
  • portfolio.md
  • people.md
  • principles.md
  • narrative.md
  • documents-presentations.md

Maintenance

  • Turn repeated playbooks into skills or automations.
  • Track posted/ready/backlog state for content.

Backend, Cloud, And Operations

ID: backend-cloud-ops

Expose service, cloud, data, auth, observability, payment, and safe external-operation routes.

Load when: Backend/API, database, cloud provider, ops, service CLI, or production maintenance work.

Read first

Skill routes

  • service-cli-registry
  • supabase
  • tanstack
  • firebase
  • azure-prepare
  • wrangler
  • stripe-webhook-hardening

Tool routes

  • Vercel
  • Cloudflare
  • Azure
  • Firebase
  • Supabase
  • Supabase TanStack DB
  • Stripe
  • Google Workspace

Generated packs

  • nextjs.md
  • backend.md
  • databases.md
  • auth.md
  • infra.md
  • cloud.md
  • observability.md
  • payments.md
  • data.md
  • ops.md
  • service-cli.md
  • google-workspace.md

Maintenance

  • Use read/list/dry-run before writes.
  • Record provider versions, URLs, IDs, and rollback paths.

Timeline

  • 2026-06-29 | Generated after Kevin requested a full capsule overhaul where capsules become the public interface to the wiki brain and attach directly to skills, tools, bookmarks, and agent packs. Source: User request, 2026-06-29