Full catalog of all notes, grouped by folder. Last updated: 2026-04-20
- Claude dispatch workflows and async AI orchestration from mobile - Orchestrate 60+ parallel AI sessions from your phone; knowledge layer compounds across surfaces
- Complete guide to Claude Code features workflows and ecosystem - Practitioner's guide: agentic loop, CLAUDE.md under 200 lines, Sonnet for 90% of tasks
- Grand unified theory of the AI hype cycle - Seven decades of AI hype follow the same arc: novel mechanism gets labeled AI, boom, bust, rename
- How LLM agents do web research: the ReAct loop - Agent research is a ReAct loop, not an algorithm; biggest failure is over-weighting official sources
- LLM agent memory synthesis April 2026 - Synthesis: 5-stage pipeline + 3 battlegrounds + harness hooks form one stack with a broken evaluation floor
- LLM agent memory systems landscape 2026 - Memory systems all solve a 5-stage pipeline; differentiation is in structure around LLM decisions
- LLM memory benchmarks and evaluation crisis - LoCoMo has ~99 wrong answers; no trustworthy single benchmark exists for memory systems
- LLM memory systems three competitive battlegrounds - Write/update loop absorbs 80% of innovation; all systems delegate conflict resolution to LLMs
- Memory systems as agent harness plugins - Memory integrates via two lifecycle hooks: before-turn recall, after-turn capture
- Multi-agent coding brain rot scan design - Fighter pilot scan patterns fix the brain rot of running 5+ AI agents in parallel
- Transformer internals for software engineers, FFN as graph database (LARQL) - FFN as sparse KNN lookup over ~348K "edges"; graph-DB reframe makes factual knowledge editable without retraining
- TurboQuant KV cache compression - Random rotation + two-stage quantizer cuts KV cache to 3-4 bits with unbiased inner products; ICLR 2026
- AI dev stack 8-layer model with tool evaluations - 8-layer stack model with SDD framework comparisons, tool scores, and 6-phase workflow
- AI tooling stack synthesis April 2026 - Synthesis: 3 layers wired through one rubric; growth and adoption-readiness are inversely correlated
- AutoResearch: the Karpathy loop pattern - Three-file contract (goal + artifact + frozen eval) ratchets quality via automated experiments
- ClaudeKit deep dive: session recovery, red team and gaps - ClaudeKit saves state on Stop hook, not just PreCompact; ship pipeline more complete than ours
- ClaudeKit evaluation and unique features - 50+ commands, interview-style spec gate, 4 adversarial reviewers; scored BOOKMARK (10/15)
- Code graph context tools for token reduction - AST-based code graphs cut agent token usage 40-95% by replacing grep with structured queries
- Context Hub vs Context7 vs the context layer ecosystem - Context Hub = curated API docs with feedback loop; Context7 = 9k+ library docs via MCP
- deepagents vs OpenClaw vs Hermes: category positioning - Library vs runtime distinction; deepagents stacks under runtimes, doesn't compete with them
- Hermes Agent comprehensive briefing April 2026 - Nous Research's self-hosted agent with auto-generated skills; 0 to 95.6K stars in seven weeks
- Hermes vs OpenClaw competitive scene April 2026 - OpenClaw wins on metrics, Hermes wins the narrative; realistic equilibrium is to run both
- OpenClaw multi-persona dev team setup playbook - End-to-end JSON5 config + SOUL/AGENTS/TOOLS files for a Telegram-led PM/Engineer/QA team
- OpenClaw virtual company pattern - "CEO/CTO/PM" multi-agent idiom is a convention, not a feature; six failure modes most writeups skip
- Prompt improvement as a learning technique - Sharpening vague prompts into structured ones is a thinking tool, not just better answers
- Tool evaluation 5-question rubric - 5 questions in 10 min; the kill question: what past failure would this have prevented?
- Why developers migrate to Hermes, ranked real vs hype - Push factor (OpenClaw CVEs + subscription cliff) beats pull factor; steal the auto-skill pattern
- Claude Code ecosystem repo evaluations for kit building - Evaluated 7 repos; methodology + hardening tools don't overlap, integration gap is the opportunity
- Claude Code hook lifecycle and event system - 21 hook events with exit code 2 as the only real enforcement; hooks beat CLAUDE.md rules
- Claude Code hook schema decision values per event type - Stop hooks need "approve"/"block", not "allow"/"deny"; mixing causes silent validation errors
- Commands vs hooks vs skills decision framework - If skipping it causes irreversible damage, use a hook; if output degrades, use a skill
- Compaction defense patterns for Claude Code sessions - Two-layer defense: PreCompact backup + post-compaction re-injection of critical rules
- Asynchronous Byzantine Fault Tolerance - aBFT removes timing assumptions; strongest fault tolerance model for permissionless blockchains
- The Bitcoin investment paradox - Either Bitcoin becomes universal (you're auto-invested) or it fails (you lose nothing)
- Cobie on (3,3) and crypto incentives - Time horizon determines VC behavior; mercenary capital erodes trust in bull markets
- Double spending in cryptocurrency - The fundamental digital currency problem: race attacks, 51% attacks, and confirmation defenses
- Ethereum token standards and security tokens - From ERC-20 to security tokens; STOs bring SEC-compliant equity and dividends on-chain
- Ray Dalio on Bitcoin as digital gold - Dalio's 2021 assessment: Bitcoin is "one hell of an invention" but risks remain
- Runtime verification for blockchain security - K Framework proves smart contract correctness mathematically; stronger than manual audits
- Stellar vs Nano comparison - Both fast and cheap; Nano is currency-only, Stellar is a platform with tokens and DEX
- Stripe on Bitcoin as the IP layer of payments - Stripe backed then dropped Bitcoin; vision of global payment protocol hit fee and speed walls
- Token emission models - Six models: halving, exponential decay, linear decay, fixed supply, constant, bonding curve
- Undercollateralized loans in DeFi - Eight approaches from flash loans to social trust pools; each trades off risk differently
- A brief totally accurate history of programming languages - Satirical timeline of programming language creation from 1800 punch cards to modern languages
- Comparing algorithm textbooks: CLRS, Tardos, Skiena, Sedgewick - Four major algo textbooks compared by use case: reference, network flow, interviews, coursework
- Go To statement considered harmful - Dijkstra's 1968 argument: go to destroys the coordinate system needed to reason about programs
- History of regular expressions - From 1950s neuroscience (Kleene's algebra) through UNIX tooling to the Russian internet outage
- History of software - resource collection - Curated links covering software history: timelines, shareware, accessibility, word processing
- The immutability of math for programmers - Math underpins CS at its core; frameworks change, linear algebra and probability do not
- Syntactic sugar, salt, and saccharin - Sugar helps readability, salt enforces discipline, saccharin is empty calories in language design
- A TCP/IP tutorial - RFC 1180 - RFC 1180 walkthrough of TCP/IP layers, ARP resolution, and packet flow between networks
- The actor model in 10 minutes - Actors replace shared memory with isolated units communicating via async message passing
- The next century of computing - End of Moore's Law triggers a Cambrian Explosion of bizarre hardware beyond 1940s architectures
- Turing completeness - A system needs branching, looping, and unbounded memory to compute anything
- What's next in computing - Computing eras cycle every 10-15 years; 2016 prediction: hardware + AI is the next wave
- Why interviewers ask linked list questions - Linked list interviews are a cultural artifact of 1980s C pointer manipulation, not a timeless test
- Why Vim uses hjkl for navigation - Chain of accidents from 1967 ASCII table to ADM-3A terminal to Bill Joy's vi
- Starship prompt configuration best practices - Start from a preset, use $fill for right-alignment, disable 90% of modules
- XDG base directory specification - XDG separates config/data/state/cache into standard dirs; simplifies dotfile management
- Vietnamese diaspora synthesis - Synthesis: structural gap (institutions, not culture) drives hollowing; bridge-builder model is the fix
- Four Asian diasporas in 2055 projected trajectories - Four variables predict diaspora survival by 2055; Vietnamese at crossroads, Chinese endure
- The bridge-builder model - Bicultural fluency connecting Vietnam's rising economy to global capital beats enclave and integration
- Vietnamese vs Chinese diaspora: structural analysis - Five structural factors explain the gap; escape-through-education trap; bridge-builder alternative
- Why Little Saigons hollow out - Successful kids leave enclaves; renters not owners means no community anchor survives the exit
- Why Vietnamese built nail salons instead of trade empires - Tippi Hedren's manicurist created a refugee entry point; subsistence businesses rational to escape
- Building Dwarves Kit from extracted patterns - Cherry-picked battle-tested patterns from 6+ repos; synthesize, don't originate
- Dwarves Kit design philosophy and architecture - 7 principles: guardrails over guidance, bash over binaries, shallow+wide
- Dwarves Kit self-assessment against philosophy - Ran /kit-health on itself; all 7 principles upheld, file count within budget
- Dwarves Kit v1.2 agent roster and CDP - 8 specialized agents + collaborative design protocol with 3 decision modes
- Dwarves Kit v1.2 ClaudeKit patterns adopted - Adopted session-state save hook and ship pipeline from ClaudeKit
- Dwarves Kit v1.2 five open decisions - Resolved: stay native Task tool, adopt codebase-memory-mcp, GSD v1 architecture
- Dwarves Kit v1.2 verification pipeline architecture - Verify-fix-retry loop: read-only verifier, scoped fix-agent, max 2 retries
- SDD landscape and Dwarves Kit v1.2 reference map - Single source of truth: 8-layer stack, all tool scores, full kit inventory
- SDD multi-agent verification architecture - 5 gaps found in v1.1; phases 1-2 built, parallel agent teams deferred
Reorganized into 6 sub-folders on 2026-04-19 (was 107 notes flat). See engineering/README.md for the sub-folder taxonomy.
- 10 tips to improve application performance - NGINX guide: reverse proxy, load balancing, caching, compression, and HTTP/2 for 10x gains
- Apache ZooKeeper for distributed coordination - Znodes, watches, sessions for leader election, service discovery, and distributed locks
- Benefits of continuous delivery - Small deploys mean lower risk, fresher context, faster feedback, and features reaching users sooner
- Creating a microservice - answer these 10 questions first - Operational checklist: testing, config, security, discovery, scaling, failure handling, upgrades, monitoring
- CSS architecture - first steps - BEM, SMACSS, ITCSS methodologies for maintainable stylesheets
- DevOps team topologies - Anti-patterns (silos, fake DevOps teams) vs recommended topologies for Dev-Ops collaboration
- Hidden dividends of microservices - Beyond scaling: microservices force explicit interfaces, independent deploys, and team autonomy
- The history of Hadoop - From Doug Cutting's Lucene to Apache Hadoop, driven by two Google papers (GFS and MapReduce)
- HTTP caching guide - Cache-Control, ETag, Last-Modified headers explained; caching layers from browser to CDN
- Testing strategies in a microservice architecture - Five-layer test pyramid for microservices: unit, integration, component, contract, end-to-end
- Monorepo advantages - Single repo for all code: atomic commits, unified tooling, easier refactoring across boundaries
- The SaaS CTO security checklist - Comprehensive checklist: infrastructure, application, and organizational security for SaaS
- Martin Fowler's software architecture guide - Architecture as shared understanding; quality drives speed; social boundaries shape systems
- The SRE model - Google's SRE: voluntary support that scales with product maturity, not a new ops title
- The junior programmer's guide to asking for help - Timebox your struggle, then ask; balance between pestering and spinning your wheels
- Chin thoi quen xau can bo neu muon theo nganh CNTT - Nine bad habits to drop for IT careers: not reading docs, skimming, copying without thinking
- Discipline doesn't scale - Calls for discipline fail because there is no motivation to adopt them; change the environment instead
- Egoless Engineering - Ego and parochialism destroy orgs; domain experts who teach beat domain owners who gatekeep
- Five problem-solving skills for software engineers - Break problems down, stay calm, think before coding, ask for help, practice regularly
- Heisenberg developers - Measuring developer work alters their behavior; estimation demands kill creativity and drive talent away
- How to learn software design - Read other people's code, study patterns, then practice by rewriting existing programs
- How to read research papers - Three-pass method: skim structure, grasp arguments, then verify details
- How to succeed as a poor programmer - Compensate for weak coding with communication, testing, simplicity, and asking for help
- Lessons from a senior developer - Hard-won lessons: ego kills growth, code reviews teach more than coding, shipping beats perfection
- Lessons learned in software development - Collected wisdom on development practices, debugging, team dynamics, and project management
- Leveraging poor memory in engineering - Poor memory forces good habits: write everything down, automate, document decisions
- Programmer competency matrix - 0-3 scale across 20+ skill categories for self-assessment; knowledge is cumulative per level
- The purple developer - 10x productivity is contextual - The 10x engineer is the one who built the system; spread the knowledge, spread the productivity
- What makes a senior developer - Seniority is judgment, positive impact beyond code, and being a force multiplier for the team
- The ACM/IEEE Software Engineering code of ethics - Eight principles covering public interest, client duties, product quality, and professional judgment
- Software engineering vs computer science - CS focuses on algorithms; SE focuses on process management for complex software systems
- 12 years, 12 lessons at ThoughtWorks - Patrick Kua on tools vs thinking, agile transformations, safety for learning, and coding architects
- Working as a software developer - Production software realities: programs are big, never done, and reading matters more than writing
- Best practices for agile documentation - Prefer executable specs, document stable concepts, keep it simple; fewer docs done well
- The antipattern scripting language - Antipatterns are contextual; in throwaway scripts, they become good ideas that speed completion
- Code for readability - Code as if the maintainer is a violent psychopath who knows where you live
- Code review basics - Fundamentals of starting code review as a team practice; overcoming resistance and building habit
- What to know before debating type systems - Static vs dynamic, strong vs weak, nominal vs structural; most type debates use terms imprecisely
- Deleting code - Delete unused code permanently; version control is your safety net, not commented-out blocks
- Effective code reviews without wasting time - One reviewer finds half of defects; beyond two reviewers, social loafing kicks in
- Mastering programming - Kent Beck's practices: slicing problems, one thing at a time, concrete then abstract
- No primitives - model domain concepts with types - Primitive Obsession code smell; wrap domain concepts in types to enforce invariants at construction
- Programming practices - Unix philosophy and beyond - Timeless principles: prototype first, fail noisily, separate policy from mechanism, least surprise
- 10 modern software over-engineering mistakes - Anticipating futures, premature abstraction, shallow wrappers, and metrics over correctness
- What makes code Swifty - Three pillars: strong type safety, path to performance, clear expressive naming
- Type wars - static vs dynamic typing history - Uncle Bob traces six decades of the type debate from Frege through Fortran to modern languages
- Write code that is easy to delete, not easy to extend - Lines of code are lines spent; build disposable software, not reusable software
- How to write a successful conference proposal - Proposals target reviewers, not audiences; the talk and the proposal are different skills
- Writing good commit messages - 50-char summary, 72-char body wrapping; explain why, not what
- Writing great documentation for open source - Start with empathy, linearize non-linear concepts, write a TOC of reader questions first
- Writing perfect pull requests - Provide context, be explicit about feedback needs, ask questions rather than issue commands
- Building web apps with functional programming - Full FP stack: Elm frontend, Haskell backend, NixOS infra for reproducible builds
- Elm language overview - Functional language compiling to JS; Model-View-Update pattern influenced Redux
- Functional programming for the rest of us - FP appears difficult due to presentation, not complexity; immutability enables easy testing and debugging
- Functional thinking - Neal Ford on shifting from OO to FP mindset: composition over inheritance, Either over exceptions
- Functional vs imperative vs declarative programming - Reference card for three paradigms: imperative (how), declarative (what), functional (pure transforms)
- Good and bad Elixir - Anti-patterns: piping side effects, over-using with, hiding higher-order functions
- Goodbye, Object Oriented Programming - OOP's three pillars dismantled: banana-gorilla-jungle, diamond problem, fragile base class
- Pragmatic functional programming - Uncle Bob: FP matters beyond concurrency; immutability brings simplicity even on 4-core laptops
- Railway Oriented Programming - Two-track error handling: happy path and failure path via Either/Result types, no monad jargon
- What is functional programming - FP defined by side effects management: pure functions with declared inputs and outputs only
- Which programming languages are functional - Side-effects management as the criterion; Haskell is genuine, JS is not, Clojure is 80%
- Why OO sucks - Joe Armstrong's critique - Erlang creator's four objections: binding data to functions, everything-is-object, scattered types, private state
- Between Go and Elixir - Complementary model: Elixir for orchestration and fault tolerance, Go for compute-heavy tasks
- Building a worker pool in Go - Job queue, workers, and dispatcher pattern for bounded concurrency in Go
- Channels in Golang - Channel types, buffering, nil/closed behavior, and edge cases for correct concurrent Go
- Comparing Elixir and Go - Concurrency model comparison: preemptive actors with isolated heaps vs cooperative goroutines with channels
- Effective error handling in Go - Indented flow pattern, custom error types, and idiomatic error handling practices
- Elixir concepts for Go developers - Actor model vs CSP comparison: addressable processes with mailboxes vs anonymous goroutines with channels
- Error handling in Upspin - Rob Pike's structured error type with Path, User, Op, Kind fields for rich context across boundaries
- Four days of Go - C/Erlang developer's candid evaluation: fast compilation but syntax inconsistencies and missing features
- Go best practices for production environments - SoundCloud's Go in production: single GOPATH, flat repo structure, go fmt on save
- Go best practices, six years in - Peter Bourgon's core principle: make dependencies explicit across config, testing, and design
- Go concurrency through illustrations - Visual introduction to goroutines, channels, and select using mining analogies
- Go context should go away - Michal Strba argues Go's context.Context is a poor design that pollutes every function signature
- Go performance optimization guide - Lock-free ring buffers 3x faster than channels; sync.Pool, escape analysis, and profiling tips
- Go proverbs - Rob Pike's Go proverbs: don't communicate by sharing memory, share memory by communicating
- Go, REST APIs, and pointers - Pointer fields solve the zero-value vs intentionally-empty ambiguity in PATCH requests
- Go testing principles by Dave Cheney - Table-driven tests, test behavior not implementation, use t.Helper() and t.Run()
- Go - the little language that could - Simplicity and pragmatism drove Go's rise past Swift, Scala, and Rust in language rankings
- A closer look at Go's type system - Named vs unnamed types, underlying types, and assignability rules that trip up Go developers
- Go vs Swift comparison - Side-by-side comparison: static typing, concurrency models, memory management, error handling
- Go 2 error handling draft design - Proposed check/handle keywords to reduce if-err-nil boilerplate; ultimately not accepted
- Idiomatic Go - Naming conventions, spelling, formatting, and style nuances from Go's standard library
- A million WebSockets and Go - Mail.Ru optimized 3M concurrent WebSockets from 72 GB to manageable with epoll and buffer pooling
- The generic dilemma in Go - Three approaches to generics: leave out (C), compile-time specialization (C++), boxing (Java)
- Typed nils in Go - Interface holds (type, data); nil concrete value in interface is non-nil, breaking nil checks
- Understanding nil in Go - Nil is the zero value for 6 types; each behaves differently when nil, enabling idiomatic patterns
- Why Go is a poorly designed language - Seven design flaws: nil interface paradox, variable shadowing, slice pain, compiler rigidity
- The Zen of Go - Dave Cheney's Go principles: single-purpose packages, flat control flow, goroutine discipline
- Bit Twiddling Hacks - Stanford reference of bitwise tricks: branchless abs, popcount, De Bruijn log, Morton interleaving
- Choose Boring Technology - Finite innovation tokens; spend them on business problems, not infrastructure novelty
- Conway's law - System architecture mirrors org communication structure; you cannot fight it, only reshape it
- Data drives code structure - Software structure follows data structure: arrays become loops, graphs become traversals
- Intro to compilers - Compiler pipeline: lexing, parsing, AST, optimization, code generation in plain language
- Papers I like (part 1) - Fabian Giesen's 10 essential CS papers: Lamport, Herlihy, Cook, and more
- Rob Pike's 5 rules of programming - Measure before optimizing; fancy algorithms are slow when n is small; data dominates
- Rust is not a good C replacement - Drew DeVault: Rust replaces C++, not C; 15 new features/year vs C's 0.73
- Stack Overflow technical deconstruction - Nick Craver's inside look at Stack Overflow infrastructure: radical transparency and embracing failure
- Pattern matching with case let in Swift - Swift's case let for destructuring enums, optionals, and tuples with pattern matching
- Technical debt as a city metaphor - Codebase as city: rushed construction, changing requirements, and patch culture cause decay
- The 80x24 rule for code formatting - 80 chars wide, 24 lines tall per method; constraints nudge toward better design
- UML as a communication tool - UML diagrams for requirements and design communication; modeling language, not a process
- What if GitHub is the devil - curl's pragmatic take - Daniel Stenberg on why curl stays on GitHub: network effect, contingency plans, pragmatism over purity
- Why big tech companies are so slow - Feature interaction complexity grows combinatorially; slowness is math, not incompetence
- Wisdom of programming quotes - Henrik Warne's curated quotes on complexity, debugging, teams, and the nature of programming
- The Zen of Python - PEP 20: beautiful over ugly, explicit over implicit, simple over complex, readability counts
- Financial knowledge as compound information advantage - Bille Finance narrative: information compounds like capital; the gap between Tier 1 and Tier 2 is learnable
- How the bond market controls housing, stocks, and jobs - Yield seesaw sets mortgage rates, equity risk premium, and corporate refinancing costs; one chain from the 10-year
- FinceptTerminal evaluation - AGPL-3 Qt6 Bloomberg-alternative; 10/15 BOOKMARK; AGPL §13 blocks integration into any network-facing trading stack
- GeckoTerminal API evaluation - Keyless H1/H4 OHLCV for DEX pools (SPL + EVM); 14/15 ADOPT; same data as $129/mo CoinGecko Pro, free
- OpenBB evaluation - Python-first financial SDK (equities, options, macro, FRED); 11/15 BOOKMARK; crypto value thin, TradFi inflection point
- OSS trading stack survey, April 2026 - 3-category synthesis (execution frameworks / agentic AI / infra libs); Freqtrade + VectorBT canonical for semi-pro crypto; ai-hedge-fund as first vendor pilot
- Why rotating ISP IPs break Binance API keys, and how to fix it with WireGuard - Cheap VPS + WireGuard beats every bundled static-IP product for exchange whitelisting; $5/mo beats $15-500/mo alternatives
- Static outbound IP solutions for crypto trading bots, ten options compared - 10 options across 5 categories; every commercial static-IP service is Category A (rent a VPS) re-bundled with markup
- Australia's Washminster government structure - Australia blends Westminster parliament with US federalism; federal/state split complicates crisis response
- How the 2026 Strait of Hormuz crisis impacts Australia - Hormuz closure hits Australia via Asian refineries; only 29-39 days fuel reserves
- Measuring oil supply disruption severity - 2026 Hormuz removed 12 mb/d, largest oil disruption ever; 400 mb strategic reserve release
- Alkaline water health claims vs reality - Stomach acid neutralizes alkaline water instantly; premium price buys marketing
- Vitamins and longevity stack - Daily supplement stack for anti-aging: NMN, Omega-3, Magnesium, CoQ10, and 12 more with dosages
- 40 best questions to ask in an interview - High-signal questions candidates should ask employers, organized by category
- Assessing software engineering candidates - Bryan Cantrill's framework: written artifacts over pop quizzes; evaluate aptitude, motivation, values
- Company culture is who you hire, fire, and promote - Culture defined by three actions, not mission statements
- Developer Happiness Index 2021 - What makes developers happy at work; data on retention factors across regions
- A developer's guide to interviewing - Questions to evaluate employers as a developer candidate
- Don't hire 0x engineers - Against the 10x myth; build functional teams with capable people
- Facebook hiring: strengths, builders, and learners - Three factors Facebook evaluates in every candidate at every level
- How to hire - Six principles: strengths over weakness, trajectory over experience, doers over talkers
- How to hire programmers and outsourced developers - Donn Felker's 4-step hiring process with programming challenges as the great equalizer
- China as a civilization state, not a nation state - China is a 2,000-year civilization wearing nation-state clothing; continuity and order trump liberty
- Imperial examinations: how China replaced religion with meritocracy - 1,300-year exam system created cultural unity and loyal bureaucracy without a holy book
- Israel, Palestine va Jerusalem - History of the Israel-Palestine conflict from shared Abrahamic origins to modern territorial disputes
- Predictive history and the ambition of psycho-history - Prof Jiang aims to build Asimov's psycho-history: connect past, explain present, predict future
- Sinicization: how China absorbs its conquerors - Every conqueror adopted Chinese civilization because demographics and bureaucracy made it inevitable
- The Tainter trap: why complexity kills empires - Empires die from complexity overload; China survives by expecting collapse and having a reset protocol
- Compound interest levels and lifestyle progression - Six wealth levels from $200/mo saver to $200M foundation; instruments change at each threshold
- How and why I invest in startups - Fund the best people on the hardest problems; measure both LP returns and happiness
- A decade of remote work - Ten years of remote lessons: go all-in or not at all, summits matter, hire for self-discipline
- The consulting secret - ask your senior ICs what is broken - Schedule 90 min with your best IC, ask what is broken, read it back to leadership
- CTO vs VP Engineering - CTO is outward-facing (vision, customers, innovation); VP Eng is inward-facing (delivery, process, people)
- How to work out what to charge clients - Calculate your floor rate from costs and billability, then apply three markup factors
- HR evaluation as unique value measurement - Market value = Differentiation x Influence; uniqueness beats commoditized skill checklists
- In pursuit of excellence - Excellence comes from unique positioning at domain intersections, not being best at common skills
- Lam an kieu Cu Ho - Five business lessons from Ho Chi Minh's methods, applied at FPT Software by Nguyen Thanh Nam
- Leadership in the agentic era - When agents absorb execution capacity, taste + trust + context become the scarce leadership work; coordination time collapses
- How to manage people who are smarter than you - Reframe management as enabling others' success; confront insecurity, learn from your team
- Masayoshi Son and the SoftBank Vision Fund - $100B fund betting AI runs the planet; gun-senryaku connects portfolio companies in a flock
- Nguyen tac truc giac trong lanh dao - Shark Phu on "tuong" traits and Maxwell on intuition as the differentiator in leadership
- Note to new design managers - Practical guide: manage time, communicate definitively, document everything, protect craft quality
- The rise of the interim CTO - Temporary tech executive bridges the gap between coding founder and strategic tech leadership
- Steve Jobs negotiation and persuasion tactics - Pitch with passion, be brutally honest, earn respect through work ethic, disarm with charm
- Tips on working with talents - Four angles on using talent: genuinely need them, create worthy challenges, fair treatment, tolerate quirks
- Why soldiers and chefs make the best product managers - Leading without authority under pressure with imperfect info; OODA loop for shipping decisions
- Why you need engineering managers - Coordination math, calendar incompatibility, and accountability require dedicated management roles
- 100 little ideas - Morgan Housel's reference card of psychological and behavioral concepts; mental models for how the world actually works
- Always be quitting: make yourself replaceable to grow - Being indispensable is a trap; 10 practices to make yourself replaceable and unlock growth
- Average Joe - An ordinary person in denial can still push far; the value is in refusing to stop trying
- Be dispassionate about your software career - Passion is a vulnerability employers exploit; invest in skills for self-preservation, not applause
- Chon nguoi hop tac va ket giao - Ancient wisdom: 6 types to avoid in business, 7 to not befriend, 4 to keep close
- Dang Le Nguyen Vu - 9 bai hoc nhan tinh the thai - Nine life lessons from Trung Nguyen's founder: self-reflect, build inner strength, accept impermanence
- Great minds discuss ideas - Three levels of discourse: people, events, ideas; the problem is when people become the endpoint
- Hygge - the Danish concept of cosiness - Hygge is a feeling of cosiness, not a lifestyle product; Danes created it to survive dark winters
- John Vu on world class quality - Bill Gates observed: small civic behaviors reveal a country's education quality and national class
- Laziness does not exist, only unmet barriers - Situational factors predict behavior better than character; respond with curiosity, not judgment
- Learning to say no without guilt - Fear of rejection drives reflexive yes; three mental shifts and practical tactics for boundaries
- The Munger Operating System for life - Charlie Munger's life principles: deserve what you want, lifetime learning, multidisciplinary thinking
- Navagraha: nine celestial bodies in Hindu astrology - Nine planets of Hindu cosmology; seven map to weekdays, Rahu/Ketu are shadow nodes governing karma
- Pavel Durov's secrets for success - Telegram founder's principles: master what you love, read constantly, write daily, stay healthy
- Simple burnout triage - One question: can you sustain the last 2 months forever? Three response levels from crisis to thriving
- Tieu chuan cua ban la gi - High standards vs low standards in work; the difference shows in how you treat details and quality
- Time is the only real currency we have - Engineers waste time on language tribalism and premature scale; invest in automation and tool mastery instead
- To chat lanh dao kinh doanh - How a Vietnamese tech corp screens management trainees; energy and persuasion are the two key tests
- Vipassana for hackers - 10-day silent meditation course explained for rational minds; systematic self-observation as mind-hacking
- We used to just live - Technology colonized every gap where unguided thinking happened; boredom was doing cognitive work we didn't realize
- What it feels like to become poor - Lost $3M in 2008 crash, ended up at a car wash; humility came from the place he feared most
- When and how to ask for help - Timebox your struggle, then ask; balance between pestering and spinning your wheels
- Why explore space - Stuhlinger's letter - NASA scientist's 1970 defense of space spending: the microscope parable and satellites fighting hunger
- Why we lie about being retired - Retirement is an identity crisis, not just a financial event; work provides meaning most can't replace
- Working attitude principles - Seven work ethic principles and ten anti-patterns; no industry is easy money, always harvest something
- MCP tool schema caching in Claude.ai connectors - Claude.ai caches MCP schemas per session; disconnect+reconnect to force refresh
- Security gates for MCP tools that bridge private to public - Server-side security gates: context-anchored secret scan, path traversal, cost-ordered pipeline
- Pattern - Backends for Frontends (BFF) - Dedicated backend per client type; avoids API bloat from serving diverse frontends
- Redundant API pre-checks in wrapper functions - Wrapper checks file existence, then library re-checks internally; doubled API calls
- Tao Te Ching + I Ching on timing and hidden preparation - The hidden dragon still trains: wu wei is preparation during a red light, not idleness
- LLM Wiki pattern: compilation over retrieval - LLM compiles raw sources into interlinked wiki instead of re-deriving via RAG each time
- Why knowledge notes need context, not just facts - Default capture depth was TIL (shallow); changing default to Atomic Note fixed quality
- Anatomy of software frauds - Three-layer fraud architecture: unlimited scapegoats, sales-driven culture, deceptive founding
- Tap trung vao san pham - When product is broken, sales and marketing accelerate failure; fix the product first
- Tesla and GM - founders vs professional managers - Steve Blank parallels Musk/Durant (visionary founders) vs Sloan (professional management)
- Anti-patterns that destroy trust permanently - Catalog of trust-killing behaviors; gossip and info leaks are unrecoverable
- Enterprise trust ladder: vendor to strategic partner - Five rungs from cold contact to strategic partner; pilot delivery is 60% process
- How to sit at the table: the thesis - Seek opportunity access and judgment transfer from elders, not money
- The 12-month progression: deposit to partnership - Trust builds in 4 phases over 12 months; most quit by month 3
- The three gates: what elders screen for - Invisible character, value, and role tests that gatekeep inner circles
- YouTube transcript extraction from cloud containers - Node.js fetch ignores HTTPS_PROXY; fix with undici ProxyAgent for YouTube transcripts
- Zed global agent rules live in the Rules Library, not AGENTS.md - No global file escape hatch; user-scope rules live in an LMDB database and need the paperclip icon to be default