---
title: "#context articles | Hashnode"
description: "Explore 100 articles tagged #context on Hashnode. Tutorials, guides, and how-tos from the community."
url: "https://hashnode.com/tag/context"
---

Tag

# #context

100 posts · 2 followers

[Varun Kumawat](https://hashnode.com/@thisisvarun) · [3d ago](https://hashnode.com/posts/how-i-built-memory-for-my-own-coding-agent-and-the-evals-that-changed-my-mind/6afc3b149bc44f0693b827ba)

### [How I built memory for my own coding agent, and the evals that changed my mind](https://devhubcommunity.hashnode.dev/how-i-built-memory-for-my-own-coding-agent-and-the-evals-that-changed-my-mind?utm_source=hashnode&utm_medium=feed)

A field report from building DoStuff, a terminal coding agent (about 25k lines of Python in the package). This is the architecture as of v0.2.15, the options I considered, and the measurements that decided between them. Where I did not measure something, I say so.

52 min read · [devhubcommunity.hashnode.dev](https://devhubcommunity.hashnode.dev/how-i-built-memory-for-my-own-coding-agent-and-the-evals-that-changed-my-mind?utm_source=hashnode&utm_medium=feed)

[Frederick Wong Sin Hin](https://hashnode.com/@fredwong76) · [Nov 3](https://hashnode.com/posts/the-great-unbundling-why-we-split-one-ai-diary-into-many/6ae961c2e37792c5d9cf93d5)

### [The Great Unbundling: Why We Split One AI Diary Into Many](https://ecupse.hashnode.dev/the-great-unbundling-why-we-split-one-ai-diary-into-many?utm_source=hashnode&utm_medium=feed)

Splitting up Agent's chat sessions to gain more details

6 min read · [ecupse.hashnode.dev](https://ecupse.hashnode.dev/the-great-unbundling-why-we-split-one-ai-diary-into-many?utm_source=hashnode&utm_medium=feed)

[thrilok e](https://hashnode.com/@thrilok) · [Nov 1](https://hashnode.com/posts/what-is-a-context-window/6ae6e8cc572bf97d5cd6f0c1)

### [What Is a Context Window?](https://thriloks-blog.hashnode.dev/what-is-a-context-window?utm_source=hashnode&utm_medium=feed)

Quick definition first. The context window of an LLM is how much text, counted in tokens, the model can look at at one time. Bigger window, longer inputs, more information going into each answer. In s

3 min read · [thriloks-blog.hashnode.dev](https://thriloks-blog.hashnode.dev/what-is-a-context-window?utm_source=hashnode&utm_medium=feed)

[Arisyn](https://hashnode.com/@arisyn) · [Oct 26](https://hashnode.com/posts/building-reliable-data-context-for-multi-step-ai-agents/6adebd08f0f8ba950cf8a9fa)

### [Building Reliable Data Context for Multi-Step AI Agents](https://arisyn.hashnode.dev/building-reliable-data-context-for-multi-step-ai-agents?utm_source=hashnode&utm_medium=feed)

Why semantic meaning, relationship evidence, and analytical grain must survive every step of an AI workflow. An enterprise AI agent can now do more than generate a SQL query. It can retrieve data, cal

7 min read · [arisyn.hashnode.dev](https://arisyn.hashnode.dev/building-reliable-data-context-for-multi-step-ai-agents?utm_source=hashnode&utm_medium=feed)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 13](https://hashnode.com/posts/repo-dump-not-context-engineering/6ace7767737b681246d5a69f)

### [Dumping the whole repo is not context engineering](https://blogs.powerduck.com/repo-dump-not-context-engineering?utm_source=hashnode&utm_medium=feed)

A bigger context window just means the model can ignore more. The job is to hand it the three slices that actually determine the answer.

3 min read · [blogs.powerduck.com](https://blogs.powerduck.com/repo-dump-not-context-engineering?utm_source=hashnode&utm_medium=feed)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 12](https://hashnode.com/posts/context-package-rots-while-agent-reads/6acd18205f0df5075aa651a9)

### [Your Agent's Context Pack Is a Snapshot. Snapshots Rot.](https://blogs.powerduck.com/context-package-rots-while-agent-reads?utm_source=hashnode&utm_medium=feed)

You hand the agent a map of the codebase. By the time it acts on it, the map is wrong. The fix isn't a bigger context window — it's cheap freshness checks.

5 min read · [blogs.powerduck.com](https://blogs.powerduck.com/context-package-rots-while-agent-reads?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/context-package-rots-while-agent-reads/6acd18205f0df5075aa651a9)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 12](https://hashnode.com/posts/agent-summary-drops-the-types/6acd41fb948e82a2cc819838)

### [When Your Agent Summarizes Your Codebase, It Quietly Drops the Types](https://blogs.powerduck.com/agent-summary-drops-the-types?utm_source=hashnode&utm_medium=feed)

An agent reads your file, writes a summary for the next session, and in that paraphrase the generic constraint, the discriminated union, the branded ID all collapse into a vague noun. The next session then builds on a version of your code that never existed.

5 min read · [blogs.powerduck.com](https://blogs.powerduck.com/agent-summary-drops-the-types?utm_source=hashnode&utm_medium=feed)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 11](https://hashnode.com/posts/treat-claude-md-like-migration-script/6acbc67a9a847cb1bcbaaf30)

### [Treat CLAUDE.md like a migration script, not a README](https://blogs.powerduck.com/treat-claude-md-like-migration-script?utm_source=hashnode&utm_medium=feed)

Instruction files used to be docs humans skimmed. Now they are runtime input an agent trusts literally. The rot that didn't matter in a README will derail your next agent run.

4 min read · [blogs.powerduck.com](https://blogs.powerduck.com/treat-claude-md-like-migration-script?utm_source=hashnode&utm_medium=feed)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 11](https://hashnode.com/posts/three-files-per-task-blast-radius-cap/6acbf0a864ebc0db7dadadf0)

### [Three files per task: the blast-radius cap I impose on every AI agent](https://blogs.powerduck.com/three-files-per-task-blast-radius-cap?utm_source=hashnode&utm_medium=feed)

The most useful guardrail I've added to my agentic workflow isn't a better model or a longer context window. It's a hard rule: three files per task. Anything bigger gets split, even when the agent insists it can handle it.

4 min read · [blogs.powerduck.com](https://blogs.powerduck.com/three-files-per-task-blast-radius-cap?utm_source=hashnode&utm_medium=feed)

[Powerduck](https://hashnode.com/@powerduck) · [Oct 10](https://hashnode.com/posts/agents-md-survives-tool-switch/6aca7e36918466d66f66f42b)

### [Your AGENTS.md Should Survive a Tool Switch](https://blogs.powerduck.com/agents-md-survives-tool-switch?utm_source=hashnode&utm_medium=feed)

The real cost of moving between Cursor, Cline, and Claude Code isn't the editor. It's re-explaining the same repo to each one from scratch. A short, tool-agnostic instructions file is the cheapest context tax you can cut.

5 min read · [blogs.powerduck.com](https://blogs.powerduck.com/agents-md-survives-tool-switch?utm_source=hashnode&utm_medium=feed)
