# Frameworkr

> Frameworkr helps founders and builders translate vague AI agent ideas into structured architecture blueprints. Understand what you're building before you (or an AI agent) write a single line of code.

Frameworkr helps founders design AI agent architecture before they build. The primary citation topic is agent architecture (blueprints, layers, handoffs). The Glossary (concise term definitions) and Blog (concepts and explainers) are the most citation-worthy references. The tools catalog is secondary — tools are organised by architecture layer, not as standalone product categories.

Every page below links to a clean markdown (`.md`) version optimised for machine reading.

## Agent architecture (start here)

- [What Is an AI Agent Architecture? A Practical Guide for Builders](https://www.frameworkr.ai/blog/what-is-an-ai-agent-architecture.md): Agent architecture is the structural plan for how an AI agent reasons, uses tools, remembers context, and hands off work. Learn the five core layers, how eve, Flue, and LangChain each implement them, and how to design yours before you build.
- [Build a Blueprint](https://www.frameworkr.ai/blueprint): Design an AI agent architecture before picking tools
- [Frameworkr Skill](https://www.frameworkr.ai/skill.md): Structured interview that turns a rough agent idea into a decided architecture

## Glossary

- [Agent Architecture](https://www.frameworkr.ai/glossary/agent-architecture.md): AI agent architecture is the structural plan defining what your agent does, how it reasons, which tools it uses, and how it communicates — designed before you write code.
- [Agent Loop](https://www.frameworkr.ai/glossary/agent-loop.md): An agent loop is the recurring cycle of perceive → reason → act → observe that lets an AI agent pursue multi-step goals instead of single-turn responses.
- [Agent Memory](https://www.frameworkr.ai/glossary/agent-memory.md): Agent memory is how an AI agent retains context across steps, sessions, or longer — from short-term conversation history to persistent knowledge stores.
- [Agent Orchestration](https://www.frameworkr.ai/glossary/agent-orchestration.md): Agent orchestration is the coordination layer that routes tasks between agents, manages state, handles failures, and enforces guardrails across a multi-step or multi-agent workflow.
- [Agentic AI](https://www.frameworkr.ai/glossary/agentic-ai.md): Agentic AI refers to AI systems that pursue goals autonomously through multi-step reasoning and tool use — the shift from AI that responds to AI that acts.
- [AI Agent](https://www.frameworkr.ai/glossary/ai-agent.md): An AI agent is software that perceives inputs, reasons about goals, and takes multi-step actions using tools — unlike rule-based automation, it exercises judgment in the moment.
- [Context Window](https://www.frameworkr.ai/glossary/context-window.md): A context window is the maximum amount of text an LLM can process in a single request — limiting how much conversation history, documents, and tool results an agent can hold.
- [Durable Execution](https://www.frameworkr.ai/glossary/durable-execution.md): Durable execution lets an AI agent pause, survive crashes or deploys, and resume mid-workflow by checkpointing each step — essential when agents wait on humans or slow systems.
- [Evals](https://www.frameworkr.ai/glossary/evals.md): Evals are scored tests that measure whether an AI agent behaves correctly — catching prompt, tool, or skill changes that quietly break quality before users do.
- [Function Calling](https://www.frameworkr.ai/glossary/function-calling.md): Function calling is the structured interface that lets LLMs invoke external functions with typed parameters — the technical foundation of agent tool use.
- [Guardrails](https://www.frameworkr.ai/glossary/guardrails.md): Guardrails are the safety boundaries, validation rules, and limits that constrain what an AI agent can do — preventing runaway loops, harmful outputs, and unauthorized actions.
- [Human-in-the-Loop](https://www.frameworkr.ai/glossary/human-in-the-loop.md): Human-in-the-loop means a person reviews, approves, or overrides an AI agent's actions at defined checkpoints — essential for high-stakes or ambiguous decisions.
- [LLM](https://www.frameworkr.ai/glossary/llm.md): A large language model (LLM) is the reasoning engine at the core of most AI agents — trained on vast text to understand, generate, and reason about language.
- [Model Context Protocol](https://www.frameworkr.ai/glossary/model-context-protocol.md): MCP is an open standard for connecting AI agents to external data sources and tools through a unified interface — like USB-C for agent integrations.
- [Multi-Agent System](https://www.frameworkr.ai/glossary/multi-agent-system.md): A multi-agent system coordinates multiple specialized AI agents — each with distinct roles and tools — to solve complex tasks that exceed a single agent's scope.
- [Prompt Engineering](https://www.frameworkr.ai/glossary/prompt-engineering.md): Prompt engineering for agents means designing system instructions, tool descriptions, and context formats that reliably steer an LLM toward correct multi-step behavior.
- [RAG](https://www.frameworkr.ai/glossary/rag.md): RAG lets AI agents retrieve relevant documents or data before generating a response — grounding answers in your knowledge base instead of model training data alone.
- [Sandbox](https://www.frameworkr.ai/glossary/sandbox.md): An agent sandbox is an isolated runtime where an AI agent can write and execute code — or otherwise act — without putting your production systems, data, or credentials at risk.
- [Skills](https://www.frameworkr.ai/glossary/skills.md): Agent skills are packaged expertise — usually Markdown playbooks — that an AI agent loads when a task needs domain knowledge, separate from the tools it uses to take action.
- [Subagent](https://www.frameworkr.ai/glossary/subagent.md): A subagent is a specialist AI agent that a parent agent delegates work to — with its own instructions, tools, and scope — so complex jobs can be split without one overloaded loop.
- [Tool Use](https://www.frameworkr.ai/glossary/tool-use.md): Tool use (function calling) lets an AI agent invoke external capabilities — APIs, databases, search, code execution — instead of relying on text generation alone.

## Blog

- [What to Ask Your Developer (or Coding Agent) About Tool Calling](https://www.frameworkr.ai/blog/what-to-ask-your-developer-about-tool-calling.md): Tool calling is where your agent stops describing and starts acting — sending the email, updating the record, moving the money. Here's what to ask before you approve that, and which answers should worry you.
- [How to Set Up an AI Agent to Ask for Reviews and Reply to Them](https://www.frameworkr.ai/blog/ai-agent-ask-for-reviews-and-reply.md): Learn how an AI agent picks the right moment to ask a customer for a review, drafts a reply worth reading, and where you still approve the response yourself.
- [How to Set Up an AI Agent to Answer "Where Is My Delivery?"](https://www.frameworkr.ai/blog/ai-agent-answer-where-is-my-delivery.md): Stop answering the same "where's my stuff?" text ten times a day. See how an AI agent reads each delivery question, checks the real status, and knows when to hand it to you.
- [How to Set Up an AI Agent to Turn Voicemails and Texts Into Booked Work](https://www.frameworkr.ai/blog/ai-agent-turn-voicemails-texts-into-jobs.md): Turn the pile of voicemails and texts you get every day into scheduled work. See what an AI agent reads before it books, where a rule fails, and where you decide.
- [How to Set Up an AI Agent to Handle Rescheduling Requests](https://www.frameworkr.ai/blog/ai-agent-handle-rescheduling-requests.md): When a customer asks to move their appointment, the right answer depends on your day. Here's how an AI agent reads a reschedule request and where you still decide.
- [How to Set Up an AI Agent to Sort Inbound Leads Worth Taking](https://www.frameworkr.ai/blog/ai-agent-sort-inbound-leads.md): Learn how an AI agent reads a new lead, decides which jobs are worth quoting and which to decline, and where you still make the call yourself.
- [How to Set Up an AI Agent to Chase Overdue Invoices Without Annoying Customers](https://www.frameworkr.ai/blog/ai-agent-chase-overdue-invoices.md): Your invoicing tool already sends reminders on a schedule. That's not an agent — and it's exactly why the reminders annoy people. Here's what an agent checks before it writes, and where the human still belongs.
- [How to Set Up an AI Agent to Follow Up on Unresponsive Quotes](https://www.frameworkr.ai/blog/ai-agent-follow-up-unresponsive-quotes.md): A reminder sequence treats every silent quote the same. Most silences aren't the same — and the difference decides whether following up wins the job or ends it.
- [How an AI Answering Service Can Transform Small Business Operations](https://www.frameworkr.ai/blog/ai-answering-service-for-small-business.md): Voicemail loses jobs and a scripted phone tree annoys callers. Here's what an AI answering service does differently when it can actually read the call — and what it must never promise on your behalf.
- [What Is LangChain? The Original Agent Framework, Explained in Plain English](https://www.frameworkr.ai/blog/what-is-langchain-agent-framework.md): LangChain defined the category in 2022, survived the backlash, and rebuilt itself around agents. Here's what the ecosystem actually looks like in 2026, who it's for, and how it compares to newcomers like eve and Flue.
- [What Is Flue? The Astro Team's Open Agent Framework, Explained in Plain English](https://www.frameworkr.ai/blog/what-is-flue-astro-agent-framework.md): Flue is an open-source, deploy-anywhere agent framework from the team behind Astro. It hit 1.0 Beta in June 2026 and has since shipped 2.0, its first stable release. Here's what it does, how it differs from Vercel's eve, and what it means for anyone scoping an agent project right now.
- [Choosing an Interaction Surface for Your Agent](https://www.frameworkr.ai/blog/choosing-an-interaction-surface-for-your-agent.md): Chat, voice, or no conversation at all — an agent's interaction surface is a design choice that propagates into the rest of the architecture. Here's how to make it deliberately.
- [How to Build a Negotiation Agent (And Why the Hard Part Isn't the Model)](https://www.frameworkr.ai/blog/how-to-build-a-negotiation-agent.md): A negotiation agent is the clearest case in agent design where the model is the easy part. Here's the loop it runs, the inputs it can't function without, where the human belongs, and the failure modes you're actually designing against.
- [Loop Engineering: The Skill Nobody Tells Non-Technical Founders About](https://www.frameworkr.ai/blog/loop-engineering.md): The prompt gets all the attention, but the loop — think, act, observe, decide — is where an agent either does the job or doesn't. Here's how to engineer it in plain English, before any code exists.
- [What Is eve? Vercel's Agent Framework, Explained in Plain English](https://www.frameworkr.ai/blog/what-is-eve-vercels-agent-framework.md): Vercel released eve in June 2026 — an open-source framework it calls "Next.js for agents." Here's what it actually does, who it's for, and what it means if you're designing an AI agent right now.
- [What Is an AI Agent? A Builder's Guide (2026)](https://www.frameworkr.ai/blog/what-is-an-ai-agent.md): Not a definition — a walkthrough. See how an AI agent handles what automation can't: the messy, ambiguous, judgment-required moments that break every workflow you've ever built.
- [Building Agent-Ready Business Systems: What Most Teams Get Wrong](https://www.frameworkr.ai/blog/building-agent-ready-business-systems.md): Most business documentation is written for humans. Here's how to build systems that work today and can be handed off to AI agents tomorrow — without rebuilding everything.

## Architecture layers (tool categories)

These are layers of an agent architecture — not primary topics. Orchestration is one layer among many.

- [Orchestration](https://www.frameworkr.ai/tools/categories/orchestration.md): Frameworks that coordinate agent logic, tool use, and decision flow
- [Models](https://www.frameworkr.ai/tools/categories/models.md): LLM providers and local inference options
- [Memory & Vector](https://www.frameworkr.ai/tools/categories/memory.md): Storage layers for agent context, embeddings, and retrieval
- [Actions & Integrations](https://www.frameworkr.ai/tools/categories/actions.md): Tools that give agents the ability to act on external systems
- [Browser & Scraping](https://www.frameworkr.ai/tools/categories/browser.md): Headless browser control and web data extraction
- [Triggers & Scheduling](https://www.frameworkr.ai/tools/categories/triggers.md): Event-driven and scheduled agent execution
- [Hosting & Infra](https://www.frameworkr.ai/tools/categories/hosting.md): Where agents run — serverless, containerised, or edge
- [Databases](https://www.frameworkr.ai/tools/categories/database.md): Persistent structured storage for agent state and outputs
- [Monitoring & Observability](https://www.frameworkr.ai/tools/categories/monitoring.md): Trace, evaluate, and debug agent runs in production
- [MCP Servers](https://www.frameworkr.ai/tools/categories/mcp.md): Model Context Protocol servers that extend agent capabilities

## Tools

- [Anthropic Claude API](https://www.frameworkr.ai/tools/anthropic-claude.md): State-of-the-art frontier models built for safe, steerable, and capable agentic use
- [Atomic Agent](https://www.frameworkr.ai/tools/atomic-agent.md): A local-first AI agent that runs on your machine, with local or cloud models.
- [Browser Use](https://www.frameworkr.ai/tools/browser-use.md): Make websites accessible for AI agents. Automate tasks online with ease.
- [Browserbase](https://www.frameworkr.ai/tools/browserbase.md): Headless browser infrastructure for agents — reliable, scalable, cloud-native
- [Chroma](https://www.frameworkr.ai/tools/chroma.md): Open-source embedding database — the fast path from prototype to local vector search
- [CloakBrowser](https://www.frameworkr.ai/tools/cloakbrowser.md): Stealth Chromium that passes every bot detection test.
- [Composio](https://www.frameworkr.ai/tools/composio.md): 250+ pre-built tool integrations so agents can act on external systems out of the box
- [Context7](https://www.frameworkr.ai/tools/context7.md): MCP server that injects up-to-date library documentation into your agent's context
- [Gemini CLI](https://www.frameworkr.ai/tools/gemini-cli.md): An open-source AI agent that brings the power of Gemini directly into your terminal.
- [Graphify](https://www.frameworkr.ai/tools/graphify.md): Turn any codebase into a queryable knowledge graph.
- [Groq](https://www.frameworkr.ai/tools/groq.md): Fastest inference API for open-source models — purpose-built for speed
- [Helicone](https://www.frameworkr.ai/tools/helicone.md): Lightweight LLM proxy for logging, cost tracking, and caching with zero code changes
- [LangChain](https://www.frameworkr.ai/tools/langchain.md): The most widely-adopted framework for building LLM-powered applications and agents
- [Langfuse](https://www.frameworkr.ai/tools/langfuse.md): Open-source LLM observability — trace every agent run, score outputs, and catch regressions
- [LlamaIndex](https://www.frameworkr.ai/tools/llamaindex.md): The data framework for building LLM applications over custom knowledge sources
- [MCP Filesystem Server](https://www.frameworkr.ai/tools/mcp-filesystem.md): Anthropic's reference MCP server for reading, writing, and navigating local filesystems
- [Obscura](https://www.frameworkr.ai/tools/obscura.md): The open-source headless browser for AI agents and web scraping.
- [Ollama](https://www.frameworkr.ai/tools/ollama.md): Run open-source LLMs locally with a simple API and no cloud dependency
- [Open Multi-Agent](https://www.frameworkr.ai/tools/open-multi-agent.md): TypeScript AI agent orchestration framework with dynamic workflows.
- [OpenAI API](https://www.frameworkr.ai/tools/openai.md): The GPT model family with vision, audio, and the broadest tool-calling ecosystem
- [OpenCLI](https://www.frameworkr.ai/tools/opencli.md): Convert any website into a CLI & run Browser Use on your logged-in Chrome.
- [Pinecone](https://www.frameworkr.ai/tools/pinecone.md): Managed vector database built for production-scale semantic search and agent memory
- [Railway](https://www.frameworkr.ai/tools/railway.md): Deploy any containerised agent to production in minutes — zero infrastructure complexity
- [Supabase](https://www.frameworkr.ai/tools/supabase.md): Open-source Firebase alternative — Postgres, auth, storage, and pgvector in one platform
- [Trigger.dev](https://www.frameworkr.ai/tools/triggerdev.md): Code-native background jobs and event-driven agent execution with full observability
- [Zapier](https://www.frameworkr.ai/tools/zapier.md): No-code automation backbone connecting 7,000+ apps for agent-triggered workflows

## Key Pages

- [Glossary index](https://www.frameworkr.ai/glossary.md): AI agent term definitions
- [Blog index](https://www.frameworkr.ai/blog.md): All articles
- [Tools catalog](https://www.frameworkr.ai/tools.md): Full tool catalog
- [Architecture layers](https://www.frameworkr.ai/tools/categories): Tools grouped by agent architecture layer
- [Pricing](https://www.frameworkr.ai/pricing): Free Basic Blueprints and Architect Mode plans