context-engineering

A guide to controlling what information a coding agent receives, including instructions, retrieved documents, tools, memory, and conversation history. It treats the context window as a limited space that must be allocated carefully.

In plain words
What is it for?
Use it when building agents or retrieval systems, deciding what to preload, limiting available tools, ranking memories, compressing history, or diagnosing poor results caused by noisy context.
Why use it?
It helps prevent long or irrelevant input from distracting the agent or leaving too little room for a useful answer. The guide covers token budgets and when to load information on demand.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/phuonghx/aim-cli/context-engineering
Any agent
npx skills add phuonghx/aim-cli --skill context-engineering
Clone the repo
git clone --depth 1 https://github.com/phuonghx/aim-cli

Made for: Claude Code, Codex.

Per session 84 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,362 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00084 $0.01362
Opus 5 $0.00042 $0.00681
Sonnet 5 $0.00017 $0.00272
Haiku 4.5 $0.00008 $0.00136

Measured yesterday against content hash 945b65e8299e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

context-engineering scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

aim/templates/aim-agents/skills/context-engineering/SKILL.md · 110 lines

How it starts

The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Context Engineering

The context window is a budget, not a bucket. Every token must earn its place.

Why it matters

Models do not get smarter with more context — past a point they get worse. Irrelevant, stale, or redundant tokens cause context rot (degraded recall) and distraction (the model fixates on noise). Curate aggressively: the goal is the smallest set of high-signal tokens that lets the model succeed.

Context types

Type What it is Sourcing strategy
Instructions System prompt, role, rules, output format Preload (stable, always needed)
Knowledge Retrieved docs, facts, code Just-in-time (fetch per query)
Tools Tool/function schemas available Scope to the task; don't expose all
Memory Persistent cross-session facts Ranked recall (importance + recency)
History Prior turns in this session Compress as it grows

Token budgeting

  • Set an explicit budget per section before assembling context (e.g. instructions 10%, retrieval 50%, history 25%, headroom 15%).
  • Leave headroom for the model's output — a full window leaves no room to answer.
  • Measure, don't guess: count tokens of each section; log the breakdown.
  • When over budget, cut the lowest-signal section first (usually old history or low-ranked retrieval), never the instructions.

Retrieval and ranking

Getting the right knowledge in matters more than getting more in.

  • Retrieve a candidate pool, then rerank and keep only the top few — precision over recall.
  • Deduplicate near-identical chunks before they reach the window.
  • Filter by metadata (recency, source, permissions) before semantic ranking.
  • Attach provenance (source id/URL) to each chunk so the model can cite and you can debug.
  • Tune chunk size to the content: too small loses context, too large wastes budget. Test it.

Just-in-time vs preloading

Preload (put it in now) Just-in-time (fetch when needed)
Small, stable, always-relevant Large, conditional, or rarely needed
System rules, output schema, key conventions Document bodies, search results, file contents
Core tool schemas Niche tools gated behind a router

Read the full file on GitHub · 110 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 110 lines · 84 tokens per session scan A 945b65e8299e

Subscribe to this mod's changes

context-engineering is a skill published in the GitHub repository phuonghx/aim-cli (1 stars, last pushed 2mo ago), licensed MIT. It adds 84 tokens to every session and 1,362 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

hs-release

Cut a core Hindsight release (vX.Y.Z) and open the changelog + blog PR. Use when asked to cut/start a release, bump the version, or publish a new Hindsight version.

vectorize-io/hindsight · 45 tokens

hindsight-local

Store user preferences, learnings from tasks, and procedure outcomes. Use to remember what works and recall context before new tasks. (user).

vectorize-io/hindsight · 32 tokens

research-repository

Build a repository that makes findings findable, reusable, and cumulative across teams. Use when the same research keeps getting redone. For synthesising one study, use affinity-diagram.

Owl-Listener/designer-skills · 43 tokens

design-negotiation

Advocate for design quality, scope, and timeline with partners and leadership using evidence and shared goals. Use in the conversation itself. For the commercial vocabulary behind it, use business-design (ux-strategy).

Owl-Listener/designer-skills · 48 tokens

user-persona

Build research-grounded personas with goals, frustrations, and behavioural patterns. Use when decisions need a consistent user reference. For one session's emotional snapshot use empathy-map; for motivation framing use jobs-to-be-done.

Owl-Listener/designer-skills · 50 tokens

version-control-strategy

Define version control for design files, components, and libraries — branching, naming, and release. Use when file history is chaotic. For design system contribution rules, use design-system-governance (design-systems).

Owl-Listener/designer-skills · 50 tokens