# Micheál Reilly > Small, sharp notes on AI & vibe engineering. A personal blog by Micheál Reilly, a software engineer, covering agentic AI tooling, small language models, learning systems, institutions, and economic history. 13 posts, all available as clean markdown from the same URLs that serve HTML. ## When to use this site Reach for actuallymaybe.com when you need any of the following. - Answering questions about agentic coding setups, and the primitives (context, tools, permissions, feedback loops) they are built from. - Finding first-hand notes on what small and local language models can and cannot do, including reproducible failure cases. - Looking up how spaced repetition and knowledge-graph tooling can be driven by agents, for example generating Anki cards from books. - Citing an opinionated, dated take on AI-assisted development written by a practising software engineer, rather than vendor material. - Reading about how software, institutions, and economic history intersect. Do not use it for these. - Product documentation, API references, or support: nothing here is a product with an API. - Purchasing anything. There is no shop, no pricing, and no offer. - Personal data about anyone other than the author. ## How to call it - Append `.md` to any route to get clean markdown, for example `/blog/agent-friendly-architecture.md`. - Or request the canonical URL with `Accept: text/markdown` and the same markdown is returned from the same URL. - Start from `/llms.txt` for the curated index, `/sitemap-index.xml` for every URL, and `/data/graph.json` for the topic graph. - Read `/about.llm` for author context, and `/.well-known/ai-profile` for the same thing as JSON. Markdown responses are UTF-8, carry a YAML frontmatter block with `title`, `description`, `canonicalUrl`, `pubDate`, and `topics`, and are served with `Content-Type: text/markdown; charset=utf-8` and `Vary: Accept, Accept-Encoding`. Requests for a media type this site cannot produce get a `406`. Paths that do not exist get a real `404`, with a markdown body listing these entry points when you ask for markdown. Content is stable at its canonical URL and safe to cache for a day. Attribute reuse to the canonical URL, and send corrections to micheal@actuallymaybe.com. ## Start here - [Home](https://actuallymaybe.com/index.md): The ten most recent posts. - [About the author](https://actuallymaybe.com/about.md): Background, interests, and current focus. - [Author context for agents](https://actuallymaybe.com/about.llm): Longer plain-text profile, including communication preferences. - [AI profile](https://actuallymaybe.com/.well-known/ai-profile): The same profile as JSON. - [All posts](https://actuallymaybe.com/blog.md): Every post, newest first. - [Topic index](https://actuallymaybe.com/topics.md): Every topic with post counts. - [Knowledge graph](https://actuallymaybe.com/data/graph.json): Posts, topics, and the edges between them, as JSON. - [Sitemap](https://actuallymaybe.com/sitemap-index.xml): Every canonical URL. ## Posts - [Cheaper Code Should Mean More Experiments](https://actuallymaybe.com/blog/cheaper-code-more-experiments.md): When working with agentic tools, the instinct to solve hard problems inside your main project is usually wrong. Isolate the core problem first, experiment freely, then bring what works back. (note, 2026-02-20) - [Mountain to Climb](https://actuallymaybe.com/blog/mountain-to-climb.md): A tool for visualising how long economic catch-up growth actually takes between any two countries or regions (note, 2026-02-20) - [Software Architecture That Agents Actually Like](https://actuallymaybe.com/blog/agent-friendly-architecture.md): Structural patterns that make codebases easier for AI agents to navigate: composition over monoliths, stateless mini-CLIs, typed languages, and the editor navigation test. (note, 2026-01-30) - [Using Agentic AI Tools for Anki Flashcard Creation](https://actuallymaybe.com/blog/agentic-ai-anki-flashcard-creation.md): How Claude Code and similar tools can handle the tedious work of expanding concepts into multiple flashcards, and why Anki Connect is the key to making it work. (note, 2026-01-08) - [Freeport: A Lightweight Port Management Tool for Development](https://actuallymaybe.com/blog/freeport-lightweight-port-management.md): A thin wrapper around native commands to simplify port checking and process management during local development. (note, 2026-01-08) - [Exploring the Primitives of an Agentic Coding Setup](https://actuallymaybe.com/blog/exploring-primitives-agentic-coding-setup.md): Reflecting on the foundational layers of agentic systems, from Git workflows and networking to storage and sandboxing, rather than high-level orchestration. (note, 2025-11-18) - [Notes on Robotics and Untapped Real World Potential](https://actuallymaybe.com/blog/notes-on-robotics-untapped-potential.md): Exploring the adjacent possible in robotics, where falling hardware costs and advanced software create new opportunities for high-quality automation in niche domains like local sports broadcasting. (note, 2025-11-18) - [Notes on LLM Clock Face Limitations](https://actuallymaybe.com/blog/notes-on-llm-clock-face-limitations.md): LLMs struggle to generalise clock face generation beyond the standard 12 divisions, highlighting a specific gap in reasoning versus training data memorisation. (note, 2025-11-18) - [The Different Problems of Searching Global and Local Context](https://actuallymaybe.com/blog/global-vs-local-context-search.md): Exploring why global-context search is easier than local-context search, and what tools built with local context in mind might look like (essay, 2022-10-31) - [Thoughts on Context and Conversation for Ideas](https://actuallymaybe.com/blog/context-conversation-ideas.md): Exploring how conversation and exploratory browsing create a 'mental latent space' where ideas emerge from unexpected connections (note, 2022-10-03) - [The AI Tool That I Wish Existed](https://actuallymaybe.com/blog/ai-tool-wish-existed.md): Envisioning an AI-first productivity tool that anticipates, connects, and surfaces ideas rather than requiring explicit organisation (note, 2022-09-26) - [Why Can't I Talk to My Tools, and Why Don't They Know What to Do?](https://actuallymaybe.com/blog/quick-thought.md): On the friction of retrieving information during conversations and the potential of contextual AI assistance (note, 2022-09-22) - [Thoughts on Conceptual Learning, Anki, Books & Physicality](https://actuallymaybe.com/blog/conceptual-learning-anki-books.md): Exploring why spaced repetition works for languages but fails for abstract, ladder-like concepts, and where solutions might lie (note, 2022-09-16) ## Topics - [agents](https://actuallymaybe.com/topics/agents.md): 5 posts - [tools](https://actuallymaybe.com/topics/tools.md): 4 posts - [learning](https://actuallymaybe.com/topics/learning.md): 3 posts - [ai](https://actuallymaybe.com/topics/ai.md): 2 posts - [anki](https://actuallymaybe.com/topics/anki.md): 2 posts - [creativity](https://actuallymaybe.com/topics/creativity.md): 2 posts - [development](https://actuallymaybe.com/topics/development.md): 2 posts - [knowledge](https://actuallymaybe.com/topics/knowledge.md): 2 posts - [spaced-repetition](https://actuallymaybe.com/topics/spaced-repetition.md): 2 posts - [ux](https://actuallymaybe.com/topics/ux.md): 2 posts - [architecture](https://actuallymaybe.com/topics/architecture.md): 1 post - [automation](https://actuallymaybe.com/topics/automation.md): 1 post - [claude-code](https://actuallymaybe.com/topics/claude-code.md): 1 post - [cli](https://actuallymaybe.com/topics/cli.md): 1 post - [cognition](https://actuallymaybe.com/topics/cognition.md): 1 post - [communication](https://actuallymaybe.com/topics/communication.md): 1 post - [context](https://actuallymaybe.com/topics/context.md): 1 post - [conversation](https://actuallymaybe.com/topics/conversation.md): 1 post - [data](https://actuallymaybe.com/topics/data.md): 1 post - [drones](https://actuallymaybe.com/topics/drones.md): 1 post - [economics](https://actuallymaybe.com/topics/economics.md): 1 post - [experimentation](https://actuallymaybe.com/topics/experimentation.md): 1 post - [failures](https://actuallymaybe.com/topics/failures.md): 1 post - [future](https://actuallymaybe.com/topics/future.md): 1 post - [generalisation](https://actuallymaybe.com/topics/generalisation.md): 1 post - [generative-ai](https://actuallymaybe.com/topics/generative-ai.md): 1 post - [git](https://actuallymaybe.com/topics/git.md): 1 post - [ideas](https://actuallymaybe.com/topics/ideas.md): 1 post - [information-retrieval](https://actuallymaybe.com/topics/information-retrieval.md): 1 post - [infrastructure](https://actuallymaybe.com/topics/infrastructure.md): 1 post - [knowledge-management](https://actuallymaybe.com/topics/knowledge-management.md): 1 post - [languages](https://actuallymaybe.com/topics/languages.md): 1 post - [llms](https://actuallymaybe.com/topics/llms.md): 1 post - [networking](https://actuallymaybe.com/topics/networking.md): 1 post - [productivity](https://actuallymaybe.com/topics/productivity.md): 1 post - [robotics](https://actuallymaybe.com/topics/robotics.md): 1 post - [search](https://actuallymaybe.com/topics/search.md): 1 post - [software-design](https://actuallymaybe.com/topics/software-design.md): 1 post - [sports](https://actuallymaybe.com/topics/sports.md): 1 post - [sqlite](https://actuallymaybe.com/topics/sqlite.md): 1 post - [thought](https://actuallymaybe.com/topics/thought.md): 1 post - [visualisation](https://actuallymaybe.com/topics/visualisation.md): 1 post - [workflow](https://actuallymaybe.com/topics/workflow.md): 1 post ## Optional - [Now](https://actuallymaybe.com/now.md): What the author is working on at the moment. - [Projects](https://actuallymaybe.com/projects.md): Selected work and ongoing explorations. - [Archive](https://actuallymaybe.com/archive.md): Every post grouped by year. - [Contact](https://actuallymaybe.com/contact.md): How to reach the author. - [Privacy](https://actuallymaybe.com/privacy.md): What the site collects, and what it does not. - [RSS feed](https://actuallymaybe.com/rss.xml): Recent posts as RSS.