https://antaoalmada.dev/Exploring the architecture, governance, and craft of AI coding agents — from agent design patterns and skills to observability and supply-chain risks. With a side of high-performance .NET for good measure. 2026-07-19T00:02:15+01:00 Antão Almada https://antaoalmada.dev/ Jekyll © 2026 Antão Almada /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png 2026-07-18T00:00:00+01:00 2026-07-18T00:00:00+01:00 https://antaoalmada.dev/posts/OpenWiki-and-OKF-Agent-Wikis/ Antão Almada

Documenting a codebase has always been hard. Writing good documentation takes time, and keeping it current as the code evolves takes discipline that most teams cannot consistently sustain. The result is the familiar pattern: documentation that starts accurate, drifts, contradicts the code, and eventually gets ignored altogether. Coding agents change this. An agent can read source files, infer ...

2026-07-03T00:00:00+01:00 2026-07-12T16:04:01+01:00 https://antaoalmada.dev/posts/Code-Agent-Knowledge-Graphs/ Antão Almada

Letting an agent grep your codebase from scratch every session is expensive and error-prone. Knowledge graph tools pre-build a navigable semantic index — functions, call graphs, types, dependencies — that agents query at near-zero cost per request. This post compares Graphify, GitNexus, codebase-memory-mcp, and CodeGraph: what each builds, how it exposes queries, and which fits your workflow. ...

2026-06-29T00:00:00+01:00 2026-07-12T16:04:01+01:00 https://antaoalmada.dev/posts/From-Hero-Prompts-to-Shared-AI-Infrastructure/ Antão Almada

A hero prompt is the long, carefully tuned instruction one developer writes to make an agent behave well — once. If that knowledge lives only in one person’s head, the team has built a dependency, not infrastructure. This post argues that the useful parts of AI workflows need to be explicit, versioned, and committed with the repository, and shows how instructions, hooks, and custom agents make ...

2026-06-27T00:00:00+01:00 2026-07-12T16:04:01+01:00 https://antaoalmada.dev/posts/Prompt-Caching-for-Coding-Agents/ Antão Almada

Usage-based billing changed the economics of coding agents. It is no longer enough to ask whether an agent can solve a task. We also need to ask how many tokens it burns while doing it. In Reducing Token Usage in Code Agents, the focus was context engineering: reduce noise, retrieve less, send better inputs. This post covers the other major lever: prompt caching — what it is, how it works, and...

2026-06-24T00:00:00+01:00 2026-07-12T16:04:01+01:00 https://antaoalmada.dev/posts/Profiling-Coding-Agents-with-Troubleshoot/ Antão Almada

Every coding agent session has a cost — tokens burned, tools called, retries attempted. GitHub Copilot’s /troubleshoot command turns opaque agent logs into a readable profile: where tokens went, which tool calls failed, and what changes will make the next run cheaper. This post shows how to use it as a profiler for agent sessions. Coding agents make this problem more interesting. They produce ...