AgentGuide

Generates role-specific learning roadmaps (Algorithm Engineer vs Development Engineer) by organizing 300+ curated resources into sequential, interview-annotated learning paths. Uses numeric prefix-based directory ordering (01-theory → 02-tech-stack → 03-practice → 04-interview) to enforce pedagogical progression, with each topic tagged for job-testing relevance and role applicability. Implements resource aggregation pattern that cites external materials rather than reproducing them, enabling lightweight maintenance while preserving signal quality.

Maintains a structured comparison matrix of agent frameworks (LangGraph, CrewAI, AutoGen, etc.) with evaluation criteria covering architecture patterns, memory systems, tool-calling approaches, and production readiness. Implements a reference-aggregation pattern that indexes official documentation and research papers rather than reimplementing framework knowledge, enabling rapid updates as frameworks evolve. Includes 12-factor agent architecture principles and agent evaluation guidelines that provide decision frameworks for framework selection.

Automates conversion of Markdown documentation into a JSON index consumed by the frontend SPA. Implemented as Python scripts in scripts/ directory that parse Markdown frontmatter, extract topic hierarchies, and generate a searchable index. Enables rapid content updates without manual index maintenance, supporting the resource-aggregation pattern by keeping documentation and index in sync.

Implements a GitHub Actions workflow (.github/workflows/deploy-pages.yml) that automatically triggers resource indexing, builds the SPA, and deploys to GitHub Pages on every commit. Enables continuous deployment of documentation updates without manual build steps. Implements a fully automated pipeline from Markdown source to live website.

Indexes RAG architecture patterns, vector database options (Pinecone, Weaviate, Milvus, Chroma), and document parsing strategies through curated reference documentation and research papers. Implements a knowledge-aggregation pattern that maps RAG papers to practical implementation guides, connecting theoretical foundations (agentic RAG, GraphRAG) to production tooling. Includes document parsing best practices covering PDF extraction, chunking strategies, and metadata preservation.

Provides structured guidance on context window management, prompt engineering patterns, and token optimization strategies for agent systems. Covers context engineering principles (how to structure prompts for agents), memory system design (conversation history, episodic memory, semantic memory), and token budget allocation across multi-turn interactions. Implements a pattern-documentation approach that catalogs proven prompt structures and context management techniques from research and production systems.

Documents SFT strategies for adapting foundation models to agent tasks, including data synthesis approaches, training pipeline design, and evaluation metrics specific to agent behavior. Covers how to generate synthetic training data for agent-specific tasks (tool-calling, reasoning, planning) and how to measure fine-tuning effectiveness. Implements a reference-aggregation pattern linking SFT research papers to practical implementation considerations.

Codifies 12-factor agent architecture principles and design patterns for building production-grade agent systems. Covers agent lifecycle management, error handling, observability, sandboxing, and safety considerations. Implements a pattern-documentation approach that catalogs proven architectural decisions from production systems and research, enabling teams to avoid common pitfalls.

Aggregates 300+ interview questions organized by role type (Algorithm Engineer, Development Engineer) and topic area, with explicit annotations for how each topic is tested in interviews. Covers LLM fundamentals, coding questions, algorithm problems, and HR/career negotiation topics. Implements a curated-signal pattern that includes only questions with direct interview relevance, avoiding generic LeetCode-style problems in favor of agent-specific technical depth.

Maintains a curated index of research papers relevant to agent development, with separate collections for agent papers and agentic RAG papers. Implements a resource-aggregation pattern that links to papers with brief context on their relevance to practical agent development. Enables researchers and engineers to quickly find foundational work (ReAct, CoT, tool-calling) and cutting-edge advances (agentic RAG, GraphRAG).

Indexes open-source agent projects and official framework guides organized by complexity and use case. Implements a resource-aggregation pattern that catalogs existing projects rather than reimplementing them, enabling learners to study production code and adapt patterns. Covers agent workflows (multi-step reasoning, tool-calling chains, multi-agent coordination) with links to reference implementations.

Provides a single-page application (SPA) frontend built with HTML/JavaScript that navigates a JSON-indexed resource collection. Implements a resource-index pipeline that converts Markdown documentation into a JSON index consumed by the frontend, enabling fast client-side search and navigation without server-side computation. Uses GitHub Pages for deployment, making the entire guide accessible without infrastructure costs.