skills.rest
simpletibr
simpletibrOFFICIAL
Explore all skills by this author →
WebsiteGitHub

simplicio-compress

Compresses prose output and standing context files while preserving code, paths, and URLs byte-for-byte.

9|Updated May 8, 2026
Search Tags:#memory optimization#llm orchestration#token compression#context compaction#prose levels#byte preservation#transform guard
One-click install
npx skills add https://github.com/simpletibr/simplicio-loop-marketing --skill simplicio-compress-simpletibr
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: simplicio-compress
Source: https://github.com/simpletibr/simplicio-loop-marketing/tree/main/.claude/skills/simplicio-compress
Command: npx skills add https://github.com/simpletibr/simplicio-loop-marketing --skill simplicio-compress-simpletibr

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Verbose AI replies, worker reports, and bloated standing context files (CLAUDE.md, AGENTS.md, notes) waste tokens on every turn. This Skill cuts output and memory tokens without losing meaning, using terse prose levels and a one-time memory compaction pass that pays back across every future iteration. ## Core Features & Use Cases - Output-side prose levels: Choose between lite, full, and ultra density tiers for replies, PR bodies, and worker-to-orchestrator reports. - Byte-preservation guarantee: Code, commands, URLs, file paths, identifiers, and version tokens are never paraphrased or reflowed; safety confirmations are never compressed. - transform_guard validation: A zero-LLM, fail-closed check compares code fences, inline-code tokens, URLs, paths, and version tokens before and after compaction, discarding any artifact that loses a token. - Use Case: Compact a bloated CLAUDE.md once, keep a .original backup, and load the terse form thereafter so every orchestrator tick costs fewer input tokens. ## Quick Start Ask the AI to compress its verbose worker report to ultra level while keeping all code, paths, and URLs exactly intact.

Frequently Asked Questions about simplicio-compress

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

FAQPage Schema
How do I reduce token usage in AI agent responses?▼

Apply terse prose levels to the model's output: `lite` drops filler phrases, `full` is normal terse technical prose, and `ultra` uses telegraphic fragments for internal reports. Compression rewrites only connective prose, never code or identifiers.

How to compact CLAUDE.md or AGENTS.md memory files safely?▼

Rewrite prose-heavy standing files to terse form while preserving code, paths, URLs, and versions verbatim, then validate with transform_guard. Keep a `.original` backup and load the compact form thereafter, re-compacting only when the source materially changes.

Does text compression ever alter code or file paths?▼

No. Code, commands, error strings, URLs, file paths, identifiers, and version tokens stay byte-exact under the one inviolable rule. The transform_guard compares token sets before and after and hard-fails any compaction that loses even one occurrence.

What happens when a compaction fails the transform_guard check?▼

A failed compaction is discarded and the original is kept byte-identical. The system issues one targeted fix touching only the flagged tokens with a maximum of two retries, then aborts to the original rather than shipping a corrupted artifact.

When should I not compress AI output or documents?▼

Never compress code, config, lockfiles, secrets-adjacent text, safety confirmations, irreversible-op warnings, or order-dependent sequences. There is also no grammar-mangling level, since degraded grammar harms code review and instructions.

Related Skills

Explore More Software Engineering Skills→
izoizo

caveman-compress

Compress verbose Claude Code memory files into token-efficient prose in place.

Community
Intermediate
infantesromeroadrianinfantesromeroadrian

token-optimizer

Compress task input to token budgets and recommend Claude model tiers.

Community
Advanced
jkerdelsjkerdels
5

dew-fast

Run a compressed plan-build-verify workflow for small software tasks.

Community
Intermediate
kangarookingkangarooking
154

context-management

Manage token budgets and context windows for long-running AI conversations.

Community
Advanced