#ai-agents
Wiki 62
- 0xsid 0xsid's personal blog at 0xsid.com — short opinion posts on agentic coding workflows and developer experience
- ADK Go 1.0 Google's Agent Development Kit for Go 1.0: tracing, plugins, HITL, YAML config, A2A protocol
- Agent Memory Components (Extractor / Store / Retriever) The three-part decomposition of an agent memory system and the one hard decision at each layer
- Agent Memory Strategy Decision Tree Five-question tree for placing each category of agent information into working, semantic, episodic, or procedural memory
- Agent readiness Site-side choices that make a web property legible to AI agents — stable URLs, machine-readable endpoints, declared crawl policy, verifiable bot identity
- Agent2Agent Protocol (A2A) Google's open protocol for cross-framework, cross-language agent communication
- Agentic Coding is a Trap Lars Faye on the supervision paradox — agent oversight requires the very skills that agent use erodes, plus vendor lock-in and token cost volatility
- Agentic Coding is Burning Me Out 0xsid on decision fatigue as the next invisible friction — gacha-loop psychology, broken pacing, and the verifier-of-the-verifier regress
- Agents vs Daemons Human-initiated AI that builds things vs self-initiated AI that maintains them — roles, not tasks
- AI Agents and the Refactoring That Never Happens Getting lost in tangled code used to be what made engineers stop and refactor; agents don't get lost, so the checkpoint has to be put back on purpose
- AI-Written Code Is Still Your Code Writing code used to force understanding it; with agents writing, understanding becomes a separate cost, and shipping the code still means owning it
- brgsk (brgsk.xyz) Blog with careful writeups on agent memory and AI systems that separate cognitive-science words from real engineering
- BubbleWrap your dev env and agents Ciężarkiewicz's Bubblewrap-based isolation for LLM agents, with Nix and tmux integration
- Disposable code When code is cheap enough to write, fail at, and throw away, maintenance dread stops deterring marginal projects and the ROI threshold shifts
- dpc.pw (Dawid Ciężarkiewicz) Blog by systems/crypto maintainer on LLM coding, dev-env isolation, open-source collaboration
- Eight Years of Wanting, Three Months of Building with AI Vibe coding failure, disciplined rewrite, the addiction loop, and why design can't be delegated
- Emad Mostaque at TechBBQ: The Internet Will Go Offline (Trending Topics) Emad Mostaque's 2026 TechBBQ talk as reported by Trending Topics, on AI attacks, owning your cognition, labor losing value, and $1.50/hour robots
- Forge — one CLI for every git forge Nesbitt's Go CLI and library that presents a single unified interface over GitHub, GitLab, Gitea/Forgejo, and Bitbucket — detects the forge from the git remote, reuses existing tokens, targets both humans and coding agents
- Human-in-the-Loop Design pattern where AI agents pause for human approval on sensitive operations
- Hybrid Search Combining BM25 keyword search with embedding similarity for better retrieval
- I don't want your PRs anymore Ciężarkiewicz's post-LLM rethink of open-source collaboration — what maintainers want from contributors now that writing code is cheap
- IngoDB — AI-Native Adaptive Database Self-morphing LSM document store in Rust; reactive indexing from query patterns, Liquid AST for AI agents
- Inverted claim onboarding Onboarding pattern where the AI agent registers first and a human "claims" it via OTP; until then, the agent has a one-recipient cap
- jxnl.co Jason Liu's writing — instructor library author, AI agent workflows, structured output, RAG, knowledge work with coding agents
- Kenn Software: local-first agent tooling suite Kenn Software's five single-binary, local-first Go tools for the agentic era — kata, middleman, agentsview, msgvault, roborev
- Lars Faye Lars Faye's personal site at larsfaye.com — articles on AI in software engineering, vendor dependency, and developer-skill economics
- Limitations of Naive RAG Failure modes of naive top-K retrieval that motivate richer agent access patterns
- LLM Wiki as agent-memory substrate Karpathy's wiki pattern is being adopted as the storage layer for agent memory products — markdown you can read and grep, instead of an opaque vector store
- LLMs are breaking 20-year-old system design Zknill on how stateful agents break the load-balancer-plus-database cloud-native assumption — the missing primitive is a routable transport name, and pub/sub channels fit
- llms.txt Proposed /llms.txt convention for giving LLMs a curated markdown overview of a website
- Memorizing Session Transcripts Isn't Useful theahura found zero SWE gain from agents searching their own past transcripts, and explains why
- Memory Conflict Detection Detecting contradictory knowledge and keeping both visible instead of silently overwriting
- Memory Decay in AI Agents Memory half-life, retrieval strengthening, and consolidation for AI agents
- Memoryfields — agent memory as a file format Cal Paterson's portable agent-memory format — markdown pages plus an optional SQLite vector index in a zip — a data structure, not a pipeline
- Mintlify's ChromaFs: A Virtual Filesystem for AI Assistants How Mintlify built a virtual filesystem to replace sandboxes and RAG for their docs assistant
- Multi-agentic Software Development is a Distributed Systems Problem Multi-agent LLM coding is a distributed consensus problem subject to FLP and Byzantine impossibility
- Open and closed models are on different exponentials Nathan Lambert on coding agents proving users pay a premium for top closed models, while open models take the larger, slower diffusion market
- Reward Hacking in the Wild 3,607 reported agent misbehaviours, LLM-classified into fourteen categories, with the caveats
- Rewriting Every Syscall in a Linux Binary at Load Time Binary rewriting technique that replaces syscall instructions with INT3 traps for complete process isolation
- Rivet Blog rivet.dev — engineering writeups from the team behind the Sandbox Agent SDK, focused on agent sandboxing and microVM infrastructure
- Sandboxing AI Agents Taxonomy of approaches for constraining what an AI coding agent can do — OS isolation, network policy, syscall filtering, HTTP interception
- Skill atrophy and the supervision paradox The circularity where agent oversight requires the critical-thinking and code-reading skills that heavy agent use measurably erodes
- Slow Software: The Case for High-latency Systems Development Irene Zhang argues critical infrastructure needs deliberate development friction now that AI decoupled importance from build time
- specification.website A single-author opinionated reference catalog of modern web specs, grouped by category and tagged Required/Recommended/Optional/Avoid
- Stateful agent routing primitive The missing piece in cloud-native architecture for talking to a specific durable process — a routable transport name that isn't a server, well-fit to pub/sub channels
- Terminal Search Clients ddgr, googler and two Kagi CLIs compared on auth cost, output format, upkeep and agent use
- The Anatomy of Agent Memory Agent memory libraries borrow cognitive-science words but ship only autobiographical semantic memory
- The cheap reverse-engineering of home devices Willison on why cheap agent-written code flips the ROI of reverse-engineering home devices — the unlock is disposability, not possibility
- The Cult of Vibe Coding Is Insane Bram Cohen on why refusing to read AI-generated code is dogfooding gone cult, with his audit-discuss-execute workflow
- The Log is the Agent ActiveGraph inverts agent architecture — the event log is truth, the graph is a projection of it
- The OpenAI/Huggingface incident; how we should manage the imminent arrival of autonomous hacking too cheap to meter Joshua Saxe reads an unreleased OpenAI model hacking Hugging Face as the start of cheap autonomous hacking, and argues for diffusion over restriction
- The specification.website Checklist A curated, opinionated checklist of nine web-spec categories (foundations, SEO, a11y, security, well-known, agent readiness, performance, resilience, i18n) with Required/Recommended/Optional/Avoid tags
- Vibe Coding and Agentic Engineering Are Getting Closer Than I'd Like Willison admits the line between vibe coding and agentic engineering has collapsed in his own practice — he no longer reviews every line, even for production
- Vibe-coding a memory tool into a 130-year-old open problem Vibe-coding an agent memory tool whose contradiction checker ran into proposition identity — the open problem of when two sentences mean the same thing
- Virtual Filesystems for LLM Agents Exposing databases as UNIX filesystems so LLM agents can browse instead of chunk retrieval
- VMs Won't Contain Cyber-Capable Agents A Trail of Bits researcher gave GPT 5.6-Cyber a VM-escape CTF; it broke out three times, using disclosed bugs, an unpatched Debian library, and its own 0-days
- We Reverse-Engineered Docker Sandbox's Undocumented MicroVM API Rivet's writeup of the undocumented `/vm` HTTP API in Docker's sandboxd daemon, what it gives you, and the Sandbox Agent SDK they built on top
- What comes next with open models Nathan Lambert on why open models should stop chasing the closed frontier and build small, specialized models that closed agents call as tools
- Who manages the agents? off-policy.com's manifesto against concentrated frontier AI — make the median worker 2x, turn everyone into an agent manager, keep the agents "sovereign" (and buy the author's product)
- Why Are AI Agents Lying, Cheating and Coordinating? (Bengio) Yoshua Bengio traces agent misbehaviour to imitation plus reward-seeking, and argues patching symptoms may only select for better-hidden cheats
- Your Container Is Not a Sandbox — MicroVM Isolation in 2026 Emir Beganović's KubeCon-EU-2026 readout on the microVM ecosystem and why agentic AI is becoming microVMs' "Docker moment"
- zknill.io Zach Knill's blog — distributed systems, durable execution, agent infrastructure, the routing primitive missing from cloud-native
Toolbox 43
- acai.sh Spec-driven dev toolkit where agents write code against numbered acceptance-criteria IDs
- ActiveGraph Event-sourced Python agent runtime where the graph is a replayable fold over an append-only log
- Agent Reading Test Benchmark that scores how well coding agents read and extract facts from web documentation
- agent-skill-linter Lints agent skill bundles for spec compliance and publishing readiness, with auto-fix
- AgentMail Hosted inbox API that gives each AI agent a real email address with WebSocket push delivery
- agentmemory Cross-agent persistent memory server with hybrid retrieval, MCP tools and hooks; implements a scored, lifecycle-managed extension of the LLM Wiki pattern
- agentsview Local-first browser, search and cost tracker for sessions across 50+ coding agents
- beads CLI issue tracker giving coding agents a dependency task graph stored in version-controlled Dolt
- Charlie Labs Daemons Markdown-defined autonomous background AI processes that watch GitHub/Linear/Slack and maintain operational hygiene
- chiasmus MCP server answering reachability, dead-code and impact questions via tree-sitter plus Prolog
- claude-code-workflow-creator Claude Code skill for authoring deterministic multi-agent workflows as plain JavaScript
- claude-design-system-prompt Reverse-engineered Claude Design system prompt plus 14 skills, tuned against AI-slop aesthetics
- CodeAlmanac A codebase wiki maintained by coding agents — plain markdown in your repo, fed from agent transcripts on a schedule and gardened daily
- CrabTrap LLM-as-a-judge HTTP proxy that intercepts and policy-checks every request an AI agent makes
- ctx Local SQLite-backed context manager for Claude Code and Codex with workstreams, conversation binding, and branching
- engrim Local SQLite memory shared across coding agents: typed decision records, FTS5 plus static-embedding recall, per-agent provenance
- flint-chart Microsoft's compact charting spec for LLM agents, compiled to Vega-Lite, ECharts or Chart.js
- flue Headless TypeScript agent harness with markdown skills and an in-process virtual bash sandbox
- forge One CLI and Go library over GitHub, GitLab, Gitea, Forgejo and Bitbucket, picked from the remote
- Foundation Contract-first Go/Rust/TypeScript full-stack substrate with enforced performance planes and agent gates
- hazmat macOS containment for AI coding agents via a separate Unix user, Seatbelt sandbox and pf rules
- Hippo Memory Agent memory system with decay by default, three storage layers and consolidation runs
- impeccable Design-skill bundle plus a 25-rule deterministic linter that flags AI-generated UI clichés
- kanbots Local-first desktop kanban board where each card runs a coding agent in its own git worktree
- kastor HCL specs for AI agents, compiled to LangGraph code or reconciled Terraform-style with plan/apply
- kata Local-first issue tracker for agent work — one SQLite daemon behind a JSON CLI and a TUI
- library-skills CLI that finds agent skills bundled inside your dependencies and symlinks them into the project
- mclaude File-based locks, handoffs and mail so two Claude Code sessions can share one repo
- MemPalace Agent memory that keeps conversations verbatim in ChromaDB under a wings-and-rooms hierarchy
- mnemonik Agent memory where each entry is Ed25519-signed and hash-anchored on Solana for later audit
- Oak Version control for coding agents — lazy mounts, messageless commits, chunk-deduped large files
- OpenHuman Local-first personal agent in Rust — memory tree mirrored to an Obsidian vault, checkpointed agent graphs, 100+ OAuth integrations, enforced privacy mode
- openwarp Warp terminal fork with bring-your-own-provider AI over six native model protocols
- peon-ping Warcraft Peon voice lines as hook-driven audio notifications for Claude Code and other agents
- Pigeon Python library for signed, attenuating capability passes an agent hands a sub-agent instead of its own API key, checked at the tool call
- QUALITY.md Root-file format declaring what quality means in a project, with a CLI that evaluates against it
- roborev Post-commit hook that reviews every commit with AI agents and feeds findings back for fixes
- Sandbox Agent SDK TypeScript SDK for running coding agents inside Docker's microVM sandboxes under one interface
- sp4rk Golang SDK for multi-agent systems with ReAct loops, MCP tools and Plan & Execute orchestration
- Stash Postgres-backed agent memory that consolidates raw episodes into facts, causal links and patterns
- statewright Rust state machine that gates which tools an AI coding agent can call in each phase of its work
- superhq macOS app running coding agents in microVMs, with credentials injected host-side, never in the VM
- tilde-run Hosted transactional sandbox for AI agents — versioned ~/sandbox composed from GitHub/S3/Drive, default-deny egress, agent-scoped RBAC, atomic commit/rollback per run