#llm-agents

A trilingual (繁中 / English / 简中) learning roadmap for agentic AI: from LLM basics to multi-agent systems, with 240+ curated resources and hands-on examples. 中文 AI agent 學習地圖。

🔁 Build reliable recurring AI-agent systems: 968 resources, 22 operational patterns, 22 loop contracts, 8 runtime starters, an interactive atlas, and a structured dataset.

fak — the Fused Agent Kernel: one Go binary that turns a tool-using agent (Claude Code, Codex, Cursor, any OpenAI/Anthropic/MCP client) into a managed agent: cache-stable model traffic, context compaction + crash resume, nanosecond tool-call policy, local GGUF serving with SSD expert offload.

Fast-ASDLC: 5x TTM with AI-native Agentic SDLC. Local-LLM first, Human-in-the-loop, Spec-driven. Built on DDD, Hexagonal Architecture, C4 Model & MCP. Features Meta-agents for self-improvement, Memory Bank for context persistence, and automated 100% test coverage. Everything-as-Code & Mermaid.js centric to save context window and slash token costs.

Compose agent skills (SKILL.md / commands) from reusable bricks, with a drift-gate.

A shared knowledge vault + full software-engineering workflow for AI coding agents — run Claude Code and Codex in parallel with one memory, one plan.

A 4-stage adversarial research auditor that fetches papers from arXiv & HuggingFace, extracts claims, and uses DeepSeek-R1 to verify them against raw abstracts.A self-correcting research & paper digest pipeline powered by local LLMs & reasoning agents

Fable 5 workflow Loop; A Claude Code skill: autonomous build / audit / verify loop that runs until a machine-checkable contract passes, with ▎ deterministic anti-spin stop conditions (no-progress, retry, oscillation, budget) and on-disk human escalation.

AI 코딩 비서(Claude·Codex·Gemini)가 확인도 안 하고 '다 됐어요' 하는 걸 자동으로 막아주는 품질 안전장치 — 비개발자·바이브코더를 위한 한국어 우선 하네스.

Proof Gradient is the agent evolution protocol where every run leaves proof, every proof selects intelligence, and every selected artifact evolves the network.

An interactive roadmap for harness engineering — building the agent loop, tool layers, context engineering, memory, retrieval, eval harnesses, and the rest of the scaffolding around modern LLMs.

Turn a book or document into a modular Agent Skill pack: atomic skills, combo workflows, a router, references, and validation.

A curated, credited library of loops — repeatable AI-agent workflows for Claude Code, Cursor & Codex. Refactor until clean, test until 100%, hunt bugs until none remain.

Attack taxonomy, toolkit & defenses for persistent-memory poisoning of LLM agents. Author: Krishita Sanjay Choksi.

AI-powered video generation with Claude Code. Describe a topic and get a fully animated MP4 with voice-synced visuals, persistent diagrams, and TTS narration. No video editing or design skills needed. Explainer videos, presentations, tutorials, product demos, onboarding, courses. Open-source multi-agent fully autonomous pipeline.

An AI-native KMS and next-gen RAG prototype. Enforces strict tree-shaped index constraints and compiler-like diagnostics after every write, enabling agents to self-heal knowledge structures and eliminate garbage-in, garbage-out RAG pain points.

Tamper-evident audit logging for OpenClaw and other local AI agents: hash-chained, locally verifiable, zero cloud dependency.

Evidence router & policy engine for coding agents — enforces source choice and proof-of-use through Claude Code hooks, including routes to MCP tools. Local-first; no LLM in the hook loop.

Pre-registered benchmark: HTML dialect + whole-tree rewrite (barkup) vs JSON, granular mutation tools, and two patch dialects, for LLM agents editing typed trees. 9,600 scored runs across 4 models. Findings in REPORT.md.

Best AI Code Architect 2026: Diagram-First Agent Orchestrator with AST Editing & Atomic Commits

Behaviour trees for AI agents. Define agentic workflows in YAML, drive deterministic agent runs from a CLI works with any agent, Claude, Codex, Copilot, Gemini etc.

Self-hosted, governed-autonomy SRE platform: an LLM agent that autonomously remediates infrastructure incidents behind a fail-closed prediction gate, mechanical verdicts, and a tamper-evident ledger. The agent can't act on a belief it hasn't checked.

AI Research Agent 2026 🔍 - Autonomous Deep Web Explorer & Adversarial Synthesis

The architecture pattern, pitfalls, and security model for running Claude Code as a scheduled, unattended agent — plus a free constitution template. Written and maintained by the agent that runs on it.

📚 Personal study notes from Google's 5-Day AI Agents course — covering Antigravity, ADK 2.0, MCP, secure agent dev, and Agent Runtime deployment.

An autonomous, LLM-powered SRE agent that takes production incidents from alert to fix/pulls metrics and logs, diagnoses the root cause, runs remediation, and writes the incident report.

Tests exact agent interventions—skills, prompts, models, tools, and workflows—and publishes what changed, what failed, and what decision follows.

A disciplined 10-stage autonomous agent pipeline (Reflexion, Tree Search, ReAct) for Claude Code and AI coding agents. Lifts local LLMs to senior engineering quality.

Generalized templates for a multi-agent Claude Code pipeline

Personal Claude Code skills — self-orchestrated task execution and more.

A multi-agent AI platform leveraging a Master Orchestrator and LLMs to unify and automate corporate finance, business intelligence, sales, and HR workflows.

Agent context gate for Codex, Claude Code, Copilot, MCP, Cursor, Cline, Gemini and Windsurf repos

Odin Codex 2026: Outline-Driven AI Agent Workflows for Codex CLI

Household decision simulation through persona-enriched multi-agent negotiation. The HDSim method core.

Residential mobility and relocation decision simulation via persona-enriched multi-agent negotiation (PEMAND)

🚀 ODIN Reflector 2026 – Next-Gen Meta-Reflection Engine for Outline-Driven AI Development

Copy citations with file links & line numbers for AI agents. Paste precise note references into any LLM chat.

Specification for governance-first context memory in autonomous AI agent systems. Defines the Mandate Gate: a pre-work mandate must exist before an agent session is valid.

Integrity scenarios for agent runtimes, distilled from real fleet incidents — deterministic fault-injection checks against real SDKs

Clone-and-run CLI: an AI onboarding agent for any codebase, demonstrating LLM + RAG + MCP + Agent working together.