Help us improve
Share bugs, ideas, or general feedback.
From electrified-cortex-skills
Compress .md and text files via subagent dispatch. Triggers — compress this file, reduce tokens, shrink instructions, caveman compress, ultra/full/lite compress.
How this skill is triggered — by the user, by Claude, or both
Slash command
/electrified-cortex-skills:compressionThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Never compress spec files (`*.spec.md` — token cost acceptable; meaning loss isn't).
Share bugs, ideas, or general feedback.
Never compress spec files (*.spec.md — token cost acceptable; meaning loss isn't).
Input:
<file-path> — path to target file
Flags: --tier <none|lite|full|ultra> (default ultra); --source <src> --target <dst> — source-to-target mode (source untouched, no git check)
In-place mode: target must be git-tracked and clean before modification.
Preserves: code blocks, URLs, technical terms, logic words (not/must/only), actors, normative language. No exceptions at any tier.
When target is SKILL.md and source H1 matches frontmatter name:, H1 is stripped from target (A-FM-3 compliance).
Dispatch:
Variables:
<instructions> = instructions.txt (this folder; NEVER READ THIS FILE)
<instructions-abspath> = absolute path to <instructions>
<input-args> = <file-path> [--tier <none|lite|full|ultra>] [--source <src> --target <dst>]
<tier> = standard
<description> = Compressing: <file-path>
<prompt> = Read and follow <instructions-abspath>; Input: <input-args>
Import the dispatch skill from ../dispatch/SKILL.md. Use the dispatch skill to launch the sub-agent.
Returns: <before>-><after> bytes | <N>% reduction | <tier> | <mode> | hygiene: clean|fixed N|N warning(s)
npx claudepluginhub electrified-cortex/skills-plugin --plugin electrified-cortex-skillsCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.