Skip to content
Get packs

Context Engineering

Helps an agent set up and manage the information it sees so its output stays accurate and focused.

Source Download skill
Repository
addyosmani/agent-skills
Stars
112,668
Licence
MIT
Files
1 file

What it does

This skill teaches an AI agent how to organise the context it works from: a persistent rules file (such as CLAUDE.md) with project conventions, only the spec sections and source files relevant to the current task, concise error output instead of whole logs, and fresh sessions at natural task boundaries. It also gives guidance on handing work off between sessions through notes and repository state, and on treating instructions found inside loaded files as data rather than commands. The skill is text guidance only; it does not itself change the code.

Great for

  • Developers starting a new project with an AI coding assistant who want a good CLAUDE.md or equivalent rules file
  • Teams whose agent keeps ignoring conventions, inventing APIs, or drifting during long sessions
  • People running multi-session or automated agent workflows who need clear restart and handoff practices

Skip it if

  • Non-technical users with no codebase or agent coding workflow, since the examples assume source files, tests and git
  • Anyone expecting the skill to run tools or fix code automatically; it only shapes how the agent prepares its context

Try asking

  • “Help me write a CLAUDE.md for this React and Express project with our commands and conventions”
  • “My agent keeps hallucinating APIs in this session, how should I reset and restructure its context?”
  • “Before I start the auth feature, tell me which files and spec sections I should load”
Step-by-step guide: how to use Context EngineeringInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    context-engineering.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip context-engineering.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip context-engineering.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Keep the rules file short and specific; the skill itself stresses that always-loaded context should be focused.
  • Load only the relevant spec section and the error lines that matter, not whole documents or full logs.
  • Start a fresh session at a completed task boundary and record progress in the spec or plan so the next session can resume safely.

What's inside

  • SKILL.md

Redistributed unmodified from addyosmani/agent-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.