Compress prompt

Skill slogsdon/skills-meta-utils/skills/compress-prompt

Meta-utility skills for Claude Code: compress prompts to save tokens, delegate work to local Qwen via LM Studio.

Install

npx -y skills add slogsdon/skills-meta-utils --skill compress-prompt

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when about to write a long prompt, include verbose text in context, or when any block of text needs to be made token-efficient before use. Delegates compression to a local model via the LiteLLM proxy (ollama-agent MCP).

SKILL.md

1.7 KB, as published. Nobody here has run it

Skill: compress-prompt

Compress verbose text into a token-efficient version using a local model via the LiteLLM proxy (default alias fast-general → gemma4:e4b-mlx). Preserves meaning and intent; strips filler, redundancy, and padding.

When to Use

Before writing a long prompt to the user
Before including verbose content in context (notes, docs, logs)
When a block of text is longer than it needs to be

Steps

Identify the text to compress (from user message, clipboard, or current context)
Call mcp__ollama-agent__qwen_start (standalone) or mcp__plugin_shane-config_ollama-agent__qwen_start (plugin — use whichever is available) with:
- task: "Compress the following into a token-efficient version. Preserve all meaning, intent, and key details. Remove filler words, redundancy, and padding. Do not summarize — compress. Output only the compressed text, no commentary.\n\n[TEXT TO COMPRESS]"
- No skill or context fields needed
Loop: if status is "running", call mcp__ollama-agent__qwen_continue (or mcp__plugin_shane-config_ollama-agent__qwen_continue in plugin) with session_id; repeat until status is "done" or "error"
Return the model's result as the compressed output

Notes

Compress ≠ summarize. Summarizing loses detail. Compressing preserves it in fewer tokens.
If the input is already tight, say so rather than padding the output.
Works on prompts, notes, documentation, session logs, or any prose.

Gives 0 of the 12 instructions most prompt engineering skills give

Counted across 563 of the 626 authors here whose files we hold, read 2026-08-06

ask at most three clarifying questionsin 22 of 563, across 15 files
respond in the user input languagein 14 of 563, across 9 files
preserve the original intentin 13 of 563, across 11 files
Establish baseline metrics and collect representative examplesin 12 of 563, across 2 files
Identify failure modes and prioritize high-impact fixesin 12 of 563, across 2 files
Apply prompt and workflow improvements with measurable goalsin 12 of 563, across 2 files
Roll back quickly if quality or safety metrics regressin 12 of 563, across 2 files
validate changes with tests and roll out in controlled stagesin 12 of 563, across 2 files
generate quantitative baseline performance reportsin 12 of 563, across 2 files
create representative test scenariosin 12 of 563, across 2 files
treat prompts as codein 12 of 563, across 5 files
test prompts on diverse inputsin 12 of 563, across 8 files

Said here and by no other author read

identify the text to compress
call the local model proxy
loop until status is done or error
return the compressed output
say so if input is already tight

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.