context-compression

A method for shortening a long coding-agent session while keeping its important decisions and unfinished work. It replaces completed conversation and bulky tool output with compact summaries.

In plain words
What is it for?
Use it after large tool outputs, when a session becomes long or repetitive, when moving to a new project phase, or when you need a concise checkpoint of completed and remaining work.
Why use it?
It keeps the session focused when the context window fills up, reducing repetition and the chance that the assistant loses track of earlier decisions.

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

Made for: Claude Code, Codex.

Per session 67 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,084 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 $0.00067 $0.01084
Opus 5 $0.00034 $0.00542
Sonnet 5 $0.00013 $0.00217
Haiku 4.5 $0.00007 $0.00108

Measured yesterday against content hash 2d2110a82c8b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

context-compression 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-compression/SKILL.md · 142 lines

How it starts

The opening of the file, as written. The whole thing — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Context Compression for Long Sessions

Over a long session the context window fills with finished work, and the assistant starts repeating itself or losing the thread. The fix is to summarize what's done — preserving the decisions, discarding the transcript — so attention stays on what's still active.

Overview

Extended sessions (roughly 30+ turns) degrade: earlier work fades, suggestions repeat, decisions get forgotten. Compressing completed phases as you go keeps the window focused on live work.

Payoff: reclaiming on the order of 5,000–15,000 tokens in a long session by swapping bulky tool output for tight semantic summaries.


When to Compress

Signal Response
Past ~20 turns Consider compressing proactively
Suggestions start repeating Window is saturated — compress now
User notes "we covered this already" Compress right away
Moving into a new phase Summarize the phase you're leaving
A tool dumps 500+ lines Compact that output on the spot

Three Levels

Level 1 — Compact a Tool Output

Shrink a single noisy result down to its meaning.

Before — raw search dump (~200 lines, ~4,000 tokens):
src/auth/jwt.ts:15: import { verify } from 'jsonwebtoken'
src/auth/jwt.ts:23: export function validateToken(token: string) {
src/auth/jwt.ts:24:   try {
... (and so on)

After — the gist (~5 lines, ~100 tokens):
Searched "jwt": 8 files, 42 hits. Core: src/auth/jwt.ts (JWT logic),
src/middleware/auth.ts (guard), src/api/login.ts (issues tokens).
Validation lives at jwt.ts:23-40, error handling at 42-55, secret read from env at line 8.

Level 2 — Summarize a Phase

Collapse a whole stretch of exploration into its conclusions.

Before — full research trail (~3,000 tokens):
[turn 1] read package.json
[turn 2] read src/index.ts
[turn 3] searched "auth"
... (a dozen more exploration turns)

After — phase summary (~200 tokens):
## Research done
- Stack: Next.js 15 app, JWT-based auth
- Auth code: 8 files across src/auth, src/middleware, src/api
- Flow: login -> mint JWT -> httpOnly cookie -> verified in middleware
- Defect: src/auth/jwt.ts:45 — expiry test uses `<` where it needs `<=`
- Plan: fix the operator, add a boundary test

Read the full file on GitHub · 142 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 · 142 lines · 67 tokens per session scan A 2d2110a82c8b

Subscribe to this mod's changes

context-compression is a skill published in the GitHub repository phuonghx/aim-cli (1 stars, last pushed 2mo ago), licensed MIT. It adds 67 tokens to every session and 1,084 once invoked, about $0.0003 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