[AIF-C01] - AWS Certified AI Practitioner (AIF-C01) - Full Course What you will learn: - Build a practical understanding of this part of LLMs and generative AI - Build the foundations before jumping into advanced LLMs and generative AI material - Understand transformer and LLM concepts without treating them as black boxes - Build intuition for generative model families and what they are useful for - Understand how agents learn from rewards, policies, and interaction loops Link is in the reply 👇 ♻️ Share this with your network if you found it useful or insightful.

Your AI coding agent is only as useful as its project context. AGENTS.md Generator Skill is an Agent Skill for generating and maintaining AGENTS.md files that guide AI coding agents through a codebase. It helps you produce concise root guidance plus scoped instructions by detecting project subsystems and extracting commands from project config files. Key features: • Thin root files – generates ~30-line root files with precedence rules and global defaults • Scoped files – detects subsystems such as backend/, frontend/, internal/, and cmd/ for more targeted guidance • Command extraction – pulls commands from Makefile, package.json, composer.json, and go.mod • Multi-language templates – includes templates for Go, PHP, TypeScript, Python, and hybrid projects • Idempotent updates – preserves an existing structure while refreshing managed content It’s open-source with split licensing: MIT for code and CC-BY-SA-4.0 for skill content and documentation. Link in the reply 👇

Your agent framework shouldn’t hide the harness you need to debug. Railtracks is a Python agent framework for developers building custom agents and harnesses. It helps you assemble tool-calling loops, tools, context, controls, and a record of every run with ordinary Python objects instead of YAML, DSLs, or a black-box runtime. Key features: • Plain Python assembly – express execution order, branching, and loops with standard Python control flow • Function-based tools – turn functions into tools and connect sub-agents or MCP servers • Context and memory options – configure system messages, retrieval, todo lists, and key-value memory toolsets • Built-in controls – add call limits, timeouts, retries, locks, guardrails, and human approval steps • Local run inspection – use the visualizer to inspect agent runs and evaluations locally without signups It’s open-source (MIT license). Link in the reply 👇

AI automation is easier to trust when you can see what the agent is doing. LiveContext CE is a self-hosted AI automation platform for teams that want chat, workflows, agents, and small internal apps in one place. It helps you turn a job described in chat into a readable workflow by pairing scoped AI agents with visual automation, interfaces, and built-in data tables. Key features: • Visual workflow engine – build execution flows with branches, loops, approvals, and triggers • Scoped AI agents – set each agent’s model, tools, files, credit budget, and audit trail • Interfaces and apps – serve forms, dashboards, and approval screens from workflows • Built-in tables – let workflows and agents read, write, filter, search, and export data • Self-hosted setup – run the Community Edition with Docker Compose or its npm launcher It’s a public, source-available repo under the LiveContext Sustainable Use License 1.0. Link in the reply 👇

AI security gaps don’t stop at the model. This map helps you trace them. CoSAI Tooling is a public GitHub repository centered on the CoSAI Risk Map, a framework for identifying, analyzing, and mitigating security risks in AI systems. It helps you assess an AI system more systematically by mapping the lifecycle across data, infrastructure, models, and applications instead of treating the model as the whole security boundary. Key features: • Lifecycle coverage – organizes the map around Data, Infrastructure, Model, and Application • Four framework areas – connects system components, risks, controls, and lifecycle personas • Human + machine-readable formats – provides YAML files alongside JSON Schema files to learn from or build on • Persona-based explorer – the static GitHub Pages explorer helps users identify relevant personas, risks, and controls • Framework mappings – cross-references MITRE ATLAS, NIST AI RMF, STRIDE, and the OWASP Top 10 for LLM Applications It’s open-source (Apache License 2.0). Link in the reply 👇

Your agent can write the follow-up. But does it know who it’s for? people-context is a local-first MCP server and CLI for giving AI agents durable memory about the people in your life. It helps you prepare more relevant messages and meetings by storing relationships, past agreements, communication preferences, and follow-ups in a SQLite database your agent can query. Key features: • Local SQLite storage – keeps the data in one file on your machine, with no account or cloud required • MCP server + CLI – connect an agent through MCP or manage records directly with `pctx` • Explainable name resolution – returns candidate matches with a reason instead of silently guessing between similar names • Reviewable imports – stage data from sources such as email, calendars, vCards, LinkedIn, Outlook, and WhatsApp exports before recording it • Privacy-minded controls – audited writes, real deletion via `forget`, and operator-only gates for sensitive records It’s open-source (MIT license). Link in the reply 👇

Picking up an AI coding task shouldn’t mean reconstructing the last session from scratch. ThreadPort is an experimental local workbench for finding unfinished AI coding tasks, reviewing their evidence, and preparing a continuation with Claude or Codex. It helps you move a task between agent environments by recording observable work state in a Context Capsule: the objective, decisions, files, commands, tests, Git identity, evidence, and next action. Key features: • Context Capsule v1 – records the observable task state needed to review or hand off work • Multi-agent adapters – maps supported Claude, Codex, Cursor, and Gemini session inputs into Capsules • Continuation preview – lets you read the complete continuation text before copying a fixed command • Explicit execution boundary – does not automatically run the next action; the terminal asks for confirmation • Local-first workflow – reads local source sessions and project contents, with no account or telemetry required It’s open-source (Apache License 2.0). Link in the reply 👇

Prompting advice is everywhere. This repo gives you a map. Awesome Prompt Engineering is a curated GitHub learning resource for developers and AI practitioners working from first prompts to context engineering. It helps you avoid hopping between random tutorials by organizing prompting techniques, agent patterns, tools, and reference material into scan-friendly guides. Key features: • Learning progression – covers basic, intermediate, and advanced prompting techniques • Context engineering – explains system prompts, retrieved context, tool definitions, and memory management • Agent-building guide – includes patterns, orchestration frameworks, tools, memory, and debugging • Builder tool directory – groups resources for building, deploying, evaluating, and governing AI systems • Fast reference material – includes an AI cheat sheet and a glossary with 148 terms It’s open-source (MIT license). Link in the reply 👇

Adding AI to a Go service shouldn’t mean rebuilding the agent loop. Dive is a foundational Go library for builders creating AI agents and LLM-powered applications. It helps you build agentic CLIs or add AI to backend services by providing a provider layer, tool calling, and an agent loop with hooks. Key features: • Multi-provider LLM access – one layer for Anthropic, OpenAI, Google, Grok, OpenRouter, Mistral, and Ollama • Agent loop + hooks – handles the generate-call-repeat cycle with extension points around each step • Broad content support – images, documents, local tools, MCP tools, and structured output work across providers • Built-in developer toolkit – includes Read, Write, Edit, Glob, Grep, Bash, WebFetch, and more • Two implementation paths – use the LLM layer directly or use the agent layer for tool-calling workflows It’s open-source (Apache License 2.0 license). Link in the reply 👇

AI clients shouldn’t force you into one model provider. Thunderbolt is an open-source, cross-platform AI client for teams that want to deploy an AI client on-prem. It helps you run an AI client with your own model providers by supporting frontier, local, and on-prem models, while documenting self-hosted backend deployment. Key features: • Cross-platform clients – available for web, iOS, Android, Mac, Linux, and Windows • Model choice – compatible with frontier, local, and on-prem models • Bring-your-own provider – add API keys for OpenAI-compatible providers or use local inference with Ollama or llama.cpp • Self-hosting docs – deployment guidance covers Docker Compose and Kubernetes • Local setup path – Make commands verify tools, install dependencies, start Postgres and PowerSync, and run the app It’s open-source (Mozilla Public License 2.0 license). Link in the reply 👇