AI Engineering · Agent Infrastructure

Disciplined patterns for autonomous code.

An engineering resource for practitioners and agents: context engineering, closed-loop test harnesses, machine endpoints, and verifiable execution invariants.

New here? what this site is · start with one idea

Machine-Readable Invariants curl -s pyweb.dev/llms.txt
$ curl -s https://pyweb.dev/api/graph.json | jq '.stats'
{
  "nodes": 185,
  "architecture": "deterministic-wiki",
  "endpoints": ["/llms.txt", "/agent", "/wiki/*.md"]
}
ZERO SCRAPING OVERHEAD FLAT MARKDOWN FIRST
Evaluated, not vibes
Every page ships through a 14-column eval matrix

Content and design changes merge only when the targeted column improves and no column regresses. Deterministic graders + LLM judges, frozen fixtures, per-column gates.

98/100
14 graded columns · methodology

Curriculum Tracks

sequenced paths, not a random walk
01 Context Rot & Progressive Disclosure context-rot → progressive-disclosure → subagents-and-context-management 02 Agentic TDD & Verification red-green-tdd → tdd-with-agents → deterministic-lint-gates 03 Multi-Agent Orchestration agents-vs-workflows → subagents → multi-agent-orchestration
69
Concept Nodes
107
Entity Models
111
Primary Sources
100%
Flat .md Endpoints

No matching resources found.

Engineering Concepts & Invariants

All wiki pages →
AG-UI Protocol ↗
Open, lightweight, event-based standard connecting AI agent runtimes to user-facing frontend applications.
agentsworkflowcontext-engineering
Agent Containment and Blast Radius ↗
Capping what an agent can do — not just what it tends to do — via sandboxes, VMs, egress controls, and classifier-gated approvals layered over fallible human oversight.
agentssecurityworkflow
Agent Harness Engineering ↗
Designing the runtime infrastructure around an agent's model — sandboxes, tool contracts, progressive disclosure, and verification gates — as the primary driver of reliability.
agentscontext-engineeringsubagents
Agent-Native Infrastructure ↗
Full-stack platforms designed around machine cognition, deterministic state machines, and cryptographic agent identity rather than human developer tools.
agentscontext-engineeringsubagents
Agentic Code Quality ↗
Risk-conditioned verification and authorization controls for autonomous software development.
coding-guidelinesfeedback-loopstdd
Agentic Engineering Patterns ↗
Disciplined software engineering practices for working with autonomous coding agents that write, test, and verify code in loops.
agentsworkflowprinciple
Agentic Manual Testing ↗
Directing coding agents to actively exercise running servers, APIs, CLI one-liners, and browser automation to verify real system behaviour.
techniqueagentsfeedback-loops
Agentic Software Factory ↗
Operating model for software development where autonomous AI agents perform end-to-end delivery under harness constraints and risk-tiered human oversight.
workflowagentscoding-guidelines
Agentic Vulnerability Lifecycle ↗
The collapse of traditional security embargo windows and CVE disclosure pipelines driven by automated coding agents that synthesize working exploits within minutes of patch discussions.
securityagentsworkflow
Agents vs Workflows ↗
The degree-of-autonomy spectrum from Anthropic's Building Effective Agents: workflows run on developer-defined code paths, agents direct their own process via tools and environment feedback.
agentspatternsllm-fundamentals
AGENTS.md Specification ↗
The root project context standard for orienting AI coding agents without system prompt bloat.
agentscontext-engineeringskills
AI Coding Taxonomy & Agent Experience ↗
Core vocabulary and operational taxonomy for AI coding agents, attention dynamics, developer/agent experience (DX/AX), and execution modes.
agentscontext-engineeringcoding-guidelines
AI Engineer Role ↗
The application-layer engineer who orchestrates AI APIs, RAG, and evals — distinct from ML Engineering by the API boundary, grounded in Latent Space's 'Rise of the AI Engineer'.
llm-fundamentalsworkflowroles
Automated Eval Engineering ↗
Techniques and harnesses for constructing reproducible, containerized agent evaluations from repository code and production traces.
evaluationagentsworkflow
Build-From-Scratch Pedagogy ↗
Building the minimal toy version from scratch as the primary mechanism for deep understanding.
pedagogytechniqueagents
Claude Managed Agents ↗
Anthropic's cloud-hosted agent runtime platform decoupling reasoning loops from containerized sandbox execution.
agentscontext-engineeringworkflow
Clean Architecture ↗
Uncle Bobs dependency-inward organizing rule applied to agent systems: source dependencies point toward stable abstractions.
principleworkflow
Closed-Loop Agent Improvement ↗
End-to-end telemetry and verification architecture where production failure traces drive automated candidate PR generation and benchmark gating.
agentsfeedback-loopsworkflow
Codex Harness Architecture ↗
Architecture of OpenAI's Codex harness: Clean/Hexagonal design, JSON-RPC app-server decoupling, multi-platform sandboxing, and bounded execution.
agentscontext-engineeringsecurity
Cognitive Debt and Walkthroughs ↗
Techniques for understanding agent-generated code through structured linear walkthroughs and interactive explanations to prevent cognitive debt.
pedagogytechniqueagents
Conformance Suites as Fitness Functions ↗
Pairing agents with uncompromising deterministic test suites so they act as autonomous optimization engines instead of drifting generators.
agentsevaluationtdd
Constraint Layering ↗
agentic-patternsharness-engineeringsecurity
Context Budget Audit ↗
Measuring and cutting per-turn token overhead — /context, logging proxies, disable flags, and deny rules to kill harness bloat.
context-engineeringagentscost
Context Engineering ↗
Curating high-signal context (AGENTS.md, CONTEXT.md, ADRs) to maximize agent reasoning performance.
agentscontext-engineeringworkflow
Context Rot ↗
The progressive degradation of LLM reasoning performance as context length and irrelevant tokens accumulate.
agentscontext-engineeringanti-patterns
Cordis Framework ↗
Microkernel plugin and dependency-injection framework powering DeepSeek Harness: lifecycle management, service containers, typed event buses, reversible side-effects.
agentsworkflowprinciple
DeepSeek Harness ↗
DeepSeek's open-source agent runtime built on the Agent = Model + Harness thesis: modular, traceable execution without hardcoded core logic.
agentscontext-engineeringsubagents
Dependency Inversion Principle ↗
principle
Dependency Rule ↗
principle
Designing for Verifiability ↗
'It's hard to eval' is a product smell — artifacts hard for you to verify are hard for users too; design checkable artifacts before building evals.
evaluationworkflowfeedback-loops
Deterministic Lint Gates ↗
Rust-speed linters with error-only configs as the Tier 1 mechanical backpressure layer for agentic coding loops.
coding-guidelinesfeedback-loopsagents
Editorial Diagrams and Visual Explanations ↗
Design principles, semantic patterns, and standards for generating high-signal, self-contained visual diagrams in agentic workflows.
diagramsvisual-explanationagents
Error Analysis and Evals ↗
Qualitative inspection of production traces to discover failure modes, build domain-specific taxonomies, and derive trustworthy evaluation rubrics.
evaluationworkflowfeedback-loops
Eval Taxonomy ↗
Evals as the AI engineer's unit tests: deterministic pass/fail checks, LLM-as-judge smoke tests, and human feedback — three tiers for wrangling predictability from probabilistic systems.
evaluationllm-fundamentalsquality
Eval-Driven Development ↗
Iterative, trace-grounded engineering discipline for discovering, encoding, and continuously testing GenAI failure modes.
evaluationfeedback-loopscoding-guidelines
Evals Skills ↗
evaluationskillsagents
First-Principles Thinking ↗
Stripping complex problems to fundamental verifiable truths and rebuilding solutions upward.
principleworldviewtechnique
Five Debts of Agentic Engineering ↗
agentic-patternscode-qualityarchitecture
Generator-Evaluator Loop ↗
Generate then adversarially evaluate: a proposal loop where a separate evaluator checks each output against explicit criteria.
agentsevaluationfeedback-loops
Git for Agentic Workflows ↗
Leveraging coding agents' native git fluency to explore repo history, resolve complex merge conflicts, bisect bugs, and rewrite clean commit stories.
gitworkflowagents
Grill With Docs ↗
Stateful interactive alignment skill that captures domain vocabulary into CONTEXT.md and three-gated decisions into ADRs.
workflowagentscontext-engineering
Grilling Doctrine ↗
Relentless adversarial interview methodology to sharpen architecture and surface assumptions before coding.
workflowagentsprinciple
Handoff Artifacts ↗
Structured, self-contained markdown documents that bridge context across session boundaries and multi-agent workflows.
agentscontext-engineeringworkflow
Hexagonal Architecture ↗
principleworkflow
Hoard and Recombine ↗
Collecting verified working code snippets and proof-of-concept tools to supply as concrete reference material for coding agents.
techniqueagentsknowledge-management
Idea-to-Ship Flow ↗
End-to-end disciplined engineering workflow moving from grilled ideas to specs, tickets, and verified PRs.
workflowagentsprinciple
Karpathy's Four Guidelines ↗
Four foundational coding conduct rules for agents: Think Before Coding, Simplicity First, Surgical Changes, Goal-Driven.
coding-guidelinesagentsprinciple
LLM App Improvement Ladder ↗
Matt Pocock's ordered ladder of 17 techniques for improving an LLM-powered app — from prompt basics to fine-tuning — try the simple thing first.
llm-fundamentalspatternsquality
LLM Message Protocol ↗
The message-based conversation protocol between application and LLM: system prompts, user/assistant messages, tool calls, and tool results.
agentscontext-engineeringllm-fundamentals
LLM Wiki Pattern ↗
Compounding, interlinked markdown knowledge base pattern curated autonomously by agents.
knowledge-managementagentstechnique
Model Context Protocol Basics ↗
MCP as the standardized bridge between LLM applications and external tools/data — the client-server layer beneath tool calling.
agentstool-useprotocols
Model Provider Abstraction ↗
Why production LLM apps need a provider-neutral interface — and the real switching cost when you code directly against one vendor's API.
llm-fundamentalsarchitecturepatterns
Multi-Agent Orchestration ↗
Hierarchical agent swarms where engineers orchestrate fleets of specialized agents in parallel isolated worktrees: coder to conductor to orchestrator.
agentsworkflowsubagents
Onion Architecture ↗
principleworkflow
Progressive Disclosure ↗
Providing context and tooling to AI agents in staged layers on-demand rather than up front.
agentscontext-engineeringskills
Prompt Bloat ↗
The accumulation of redundant, generic, or conflicting instructions in system prompts that degrades agent performance.
agentscontext-engineeringanti-patterns
Red/Green TDD ↗
Test-first development where failing tests are observed before writing minimal implementation code - the core discipline for coding agents.
tddworkflowagents
Releasable Patch Rate ↗
agentic-patternscode-qualityworkflow
Site-as-Eval-Subject Grading ↗
Grading a static site with deterministic graders and LLM judges, then hill-climbing the design against the scorecard.
evaluationagentstechnique
Skill Treatment Effect ↗
agentic-patternsevaluationharness-engineering
Smart Zone ↗
The optimal token window where frontier LLMs maintain peak reasoning sharpness before attention degradation sets in.
agentscontext-engineeringworkflow
Software Engineering Fundamentals for Agents ↗
agentic-patternsarchitectureworkflow
Structured Outputs ↗
Constraining LLM responses to JSON-schema shapes — objects, enums, arrays — for data extraction and classification, with streaming and tool-based variants.
llm-fundamentalscontext-engineeringpatterns
Subagents and Context Management ↗
Preserving root conversation context by dispatching isolated sub-agents for repo exploration, testing, debugging, and file edits.
agentssubagentscontext-engineering
TDD With Agents ↗
Rigorous test-driven development loop using executable tests as the truth referee for coding agents.
tddworkflowagents
The Feynman Technique ↗
Teaching and learning methodology: plain English explanations, identifying gaps, and radical simplification.
techniqueteachingpedagogy
Think Tool ↗
Anthropic's technique of giving an LLM a no-op 'think' tool so it can persist structured reasoning into context before complex tool calls.
agentstool-usepatterns
Tool Calling Loop ↗
The four-step agentic loop — specify tools, receive tool calls, execute locally, return results — that lets LLMs act on the world.
agentstool-usellm-fundamentals
Tracer Bullets ↗
Shipping the thinnest end-to-end slice through all architecture layers before expanding features.
workflowagentstechnique

Entities & Thought Systems

Addy Osmani ↗
Engineering leader and author writing on agent harness engineering, agentic code quality gates, and software factory workflows.
personworkflow
Aditi Raghunathan ↗
person
Ahmed E. Hassan ↗
person
Airbnb ↗
Global travel marketplace and engineering pioneer in industrial Eval-Driven Development (EDD) for generative AI applications.
organizationevaluationworkflow
Alistair Cockburn ↗
personeducator
Andrei Botocan ↗
person
Andrej Karpathy ↗
AI researcher, educator, Stanford CS231n co-creator, and developer of the build-from-scratch pedagogy.
personeducatoragents
Andrew Ng ↗
personeducator
Anil Madhavapeddy ↗
personsecurity
Anthropic ↗
AI safety research company developing Claude LLM models and researching agent systems.
companyagentsevaluation
Armin Ronacher ↗
Software engineer critiquing coordination friction loss in agent-accelerated development.
personprinciple
Beatrice Casey ↗
person
Beth Barnes ↗
person
Boris Cherny ↗
Engineering leader at Anthropic, creator of Claude Code, and author of Programming TypeScript.
personauthorityworkflow
Brittany Reid ↗
person
Cathryn Lavery ↗
Designer, entrepreneur, and author of Diagram Design; pioneer of editorial visual design systems and semantic diagramming for AI coding agents.
personeducatoragents
Christoph Csallner ↗
person
Christoph Nakazawa ↗
personcoding-guidelinesfeedback-loops
CopilotKit ↗
Open-source AI Copilot framework and creators of the AG-UI protocol for agent-user interaction.
companyagentsworkflow
Cursor ↗
AI-powered code editor company pioneering hierarchical agent systems for autonomous coding.
companyagentsworkflow
David Gros ↗
person
David Rein ↗
person
DeepSeek ↗
AI research company behind frontier open-weights models and the DeepSeek Harness agent runtime developer preview.
companyagents
Dipayan Banik ↗
person
DORA ↗
companyevaluation
Drew Breunig ↗
Writer on LLM economics; argued in 'Fable & The End of the Free Lunch' that high frontier-model pricing ended the era when new models papered over weak harnesses.
personagentscontext-engineering
Dung Nguyen Manh ↗
person
Eirini Kalliamvakou ↗
person
Epoch AI ↗
companyevaluation
Florian Brand ↗
person
Gelei Deng ↗
person
Gergely Orosz ↗
Author of The Pragmatic Engineer newsletter, covering tech industry trends, engineering leadership, and developer tooling in the agentic era.
personauthorityworkflow
GitHub ↗
Leading software development platform, developer of Copilot, and author of the Spec-Driven Development toolkit for AI agents.
organizationworkflowtool-use
Hajimu Iida ↗
person
Hamel Husain ↗
AI product engineer, machine learning educator, and specialist in LLM evaluation, error analysis, and domain-grounded AI systems.
personeducatorevaluation
Hao Li ↗
person
Haoxiang Zhang ↗
person
Hiroshi Iwata ↗
person
Huy Nhat Phan ↗
person
Ivan Bercovich ↗
person
Ivar Jacobson ↗
personeducator
Ivgeni Segal ↗
person
James Coplien ↗
personeducator
James Shore ↗
Software engineering consultant analyzing economic implications of AI-accelerated development.
personprinciple
Jarred Sumner ↗
Creator of Bun, who led autonomous agent-driven rewrite of Bun's codebase from Zig to Rust.
personagentsevaluation
Jean-Stanislas Denain ↗
person
Jeffrey Palermo ↗
personeducator
Joanna C. S. Santos ↗
person
Joel Becker ↗
person
Johann Rehberger ↗
Security researcher warning about normalization of deviance in AI system permissions.
personagentsanti-patterns
K M Ferdous ↗
person
Ken’ichi Yamaguchi ↗
person
Kent Beck ↗
Pioneer of Extreme Programming (XP), Test-Driven Development (TDD), co-author of the Agile Manifesto, and author writing on AI coding workflows.
personauthoritytdd
Kexun Zhang ↗
person
Kowshik Chowdhury ↗
person
Lalit Maganti ↗
Syntaqlite creator documenting where coding agents help implementation and harm unresolved design work.
personagentsanti-patterns
Leo Yu Zhang ↗
person
Maliheh Izadi ↗
person
Mario Zechner ↗
Software engineer warning about compounding errors in high-velocity agent development.
personprinciple
Martin Fowler ↗
Author, chief scientist at ThoughtWorks, and foundational thinker on software architecture, refactoring, and AI-assisted engineering practices.
personauthorityworkflow
Matt Pocock ↗
TypeScript and AI engineering educator behind the idea-to-ship workflow and deep module design.
personeducatorworkflow
Mert Demirer ↗
person
Miku Watanabe ↗
person
Minh Vu Thai Pham ↗
person
Model Evaluation & Threat Research ↗
companyevaluation
Mohammed Latif Siddiq ↗
person
Nat Pryce ↗
personeducator
Nate Rush ↗
person
Nghi D. Q. Bui ↗
person
Nick Craig-Wood ↗
personsecurityworkflow
OpenAI ↗
AI research and deployment company developing GPT models, ChatGPT, Codex, and agent harness infrastructure.
companyagentsevaluation
Peter Cihon ↗
person
Prem Devanbu ↗
person
Prithvi Rajasekaran ↗
personagentsevaluation
Rahul Pandita ↗
person
Razvan Mihai Popescu ↗
person
Richard Feynman ↗
Nobel laureate physicist known for first-principles thinking, joyful curiosity, and plain-language pedagogy.
personphysicistteaching
Robert C. Martin (Uncle Bob) ↗
personeducator
Sabrina Haque ↗
person
Sarvesh Ingale ↗
person
Shashwat Saxena ↗
person
Shazibul Islam Shamim ↗
person
Shopify ↗
E-commerce platform company conducting autonomous agent optimization experiments.
companyagentsevaluation
Shota Sawada ↗
person
Shreya Shankar ↗
Computer science researcher at UC Berkeley focusing on data management, ML systems, and active-learning tooling for LLM error analysis and evaluation.
personeducatorevaluation
Sida Peng ↗
person
Simon Willison ↗
Software engineer, creator of Datasette, co-creator of Django, and prominent authority on LLM tooling, evals, and agentic engineering patterns.
personauthorityworkflow
Steve Freeman ↗
personeducator
Stripe ↗
Financial infrastructure platform conducting benchmark research on autonomous coding agents building production payment integrations.
organizationevaluationbenchmarks
Tatsuya Shirai ↗
person
Tessl ↗
AI software development company researching developer platforms, agentic software engineering benchmarks, and system harness architectures.
companyevaluation
Tobias Lütke ↗
Shopify CEO who conducted autonomous agent optimization of Liquid template engine.
personagentsevaluation
Trygve Reenskaug ↗
personeducator
Tue Le ↗
person
Vercel ↗
Frontend platform company; creator of the AI SDK, a provider-neutral TypeScript interface for LLM generation, streaming, and tool calling.
companyllm-infrastructure
Vinicius Carvalho Lopes ↗
person
Viv Trivedy ↗
Practitioner credited by Addy Osmani with naming harness engineering and articulating the model-plus-harness framing.
personagentscontext-engineering
Wes McKinney ↗
Creator of pandas, POSIT; pioneer of high-throughput agentic engineering harness workflows and automated post-commit review.
personworkflow
Wilson Lin ↗
Cursor researcher documenting planner-worker architectures for long-running autonomous coding.
personagentsworkflow
Xinye Zhao ↗
person
Yanjun Zhang ↗
person
Yi Liu ↗
person
Ying Zhang ↗
person
Yubin Qu ↗
person
Yuekang Li ↗
person
Yutaro Kashiwa ↗
person
Ziqian Zhong ↗
person

Architectural Guides

All posts →
2026-08-30 The Agentic Engineering Curriculum A structured course on AI engineering and agentic patterns, distilled from the wiki: foundations, context, evals, verification, harnesses, and the software factory. 2026-08-28 The Linter Is the Cheapest Adult in the Room Why Rust-speed, error-only lint gates like Oxlint are the Tier 1 backpressure layer that keeps autonomous coding loops honest — with the failure modes they mechanically close. 2026-08-27 The Agent as a Virtual OS: Inside the Codex Harness Architecture Dissecting OpenAI's Codex harness: JSON-RPC app-server decoupling, multi-platform OS sandboxing, bounded context streams, and the lessons of production agent design. 2026-08-27 The Five Debts of Agentic Engineering Why probabilistic coding generators accumulate intent, semantic, verification, architecture, and authorization debt—and how risk-conditioned systems keep autonomous loops releasable. 2026-08-25 Clean Architecture for Agent Harnesses How Uncle Bob's 2012 Dependency Rule and boundary isolation principles solve the durability, containment, and testability crises in modern AI agent harnesses. 2026-08-24 Skills-Native Evaluation and Agent Adversarial Training How evaluation methodology shipped as installable skills and GAN-inspired generator-evaluator loops are advancing AI agent reliability beyond prompt tweaks. 2026-08-24 Loop and Graph Engineering: The Dual Topologies of Agentic Systems Why single-agent prompt loops degrade past the smart zone, and how graph orchestration provides the structural partition needed for production AI systems. 2026-08-22 Hello world The first post on pyweb.dev 2026-08-22 Building Effective Agentic Workflows A practitioner's architectural guide to building reliable AI coding agents: avoiding context rot, operating in the smart zone, and enforcing tight engineering loops.