skills.rest
0xharryriddle
0xharryriddleCOMMUNITY
Explore all skills by this author →
𝕏@0xHarryNguyenVNGitHub

context-compression

Generate probes and score context compression quality against a rubric.

3|Updated Mar 22, 2026
Search Tags:#token-optimization#LLM-judge#evaluation-framework#context-compression#artifact-trail#anchored-summarization#probe-evaluation
One-click install
npx skills add https://github.com/0xharryriddle/codex-field-kit --skill context-compression-0xharryriddle
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: context-compression
Source: https://github.com/0xharryriddle/codex-field-kit/tree/main/archive/upstream/chasebuild-agent-skills/context-engineering/skills/context-compression
Command: npx skills add https://github.com/0xharryriddle/codex-field-kit --skill context-compression-0xharryriddle

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Compressing long AI-agent conversations without sacrificing essential information is a core challenge. This skill provides a structured approach to preserve artifacts, decisions, and context while reducing token usage.

Core Features & Use Cases

  • Anchored iterative summarization to maintain file paths, decisions, and intent across multiple compressions.
  • Structured evaluation framework that measures accuracy, artifact-trail, continuity, and instruction adherence.
  • Suitable for long-running sessions, code exploration, and multi-agent planning where context windows are limited.

Quick Start

Summarize only the newly truncated content while preserving file state, decisions, and intent.

Frequently Asked Questions about context-compression

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I compress context in long AI agent conversations without losing file paths and decisions?▼

To compress context without losing critical details, apply anchored iterative summarization to preserve file state, decisions, and intent across multiple truncations while reducing token usage.

What is anchored iterative summarization for multi-file projects?▼

Anchored iterative summarization is a context compression mechanism that maintains file paths, decisions, and intent across multiple compression cycles, ensuring critical artifacts remain intact for long-running sessions.

How do I evaluate the quality of context compression for LLM agents?▼

You evaluate context compression quality by generating probes and scoring agent responses against a structured rubric that measures accuracy, artifact-trail continuity, and instruction adherence.

Can I use context compression techniques for multi-agent planning scenarios?▼

Yes, context compression suits multi-agent planning where context windows are limited, providing a structured approach to minimize token usage while preserving essential artifacts and decisions.

What are the limitations of iterative context compression in long-running sessions?▼

Iterative context compression requires careful preservation of artifact trails and decisions; without proper probe evaluation, multiple compression cycles risk losing critical information and instruction adherence.

Does context compression work without a structured evaluation framework?▼

Context compression relies on a structured evaluation framework combining a probe generator and evaluator to measure accuracy and continuity, ensuring token optimization does not sacrifice essential information.

Related Skills

Explore More Software Engineering Skills→
samvanmesamvanme

context-compression

Compress long conversation histories to optimize token usage while preserving critical context.

Community
Advanced
ChakshuGautamChakshuGautam

context-compression

Compress agent session context using anchored iterative summarization and probe-based evaluation.

Community
Advanced
chef0111chef0111
1

context-engineering

Curate minimal high-signal token sets for LLM context optimization.

Community
Advanced
ken-cavanagh-gleanken-cavanagh-glean
4

context-compression

Compress long-running agent session context with anchored iterative summarization.

Community
Basic