Skills compress

📦

compress

Name: compress
Author: juliusbrussee

Content revision r1 High Risk 📁 Filesystem access⚙️ External commands🔑 Env variables

Compress Memory Files Safely

Long memory files consume context and make repeated AI sessions expensive. This skill compresses markdown and text memory files while preserving technical details, code, URLs, and structure.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "compress" from https://skillstore.io/skills/juliusbrussee-compress.md and its manifest at https://skillstore.io/api/skills/juliusbrussee-compress/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/juliusbrussee-compress.md Signed manifest GET /api/skills/juliusbrussee-compress/manifest Signed lockfile GET /api/skills/juliusbrussee-compress/lockfile

Test it

Using "compress". A verbose project instruction file with repeated guidance and examples.

Expected outcome:

A shorter markdown file that keeps the same headings, examples, commands, and links, plus a backup of the original.

Using "compress". A code or configuration file path is provided by mistake.

Expected outcome:

The skill reports that the file is not natural language and skips compression without modifying it.

Using "compress". A markdown file contains code blocks and URLs.

Expected outcome:

The prose is shortened while code blocks and URLs remain unchanged.

Security Audit

High Risk

v6 • 7/9/2026 Open versioned report

Most static findings are false positives caused by markdown backticks, validation regexes, and denylist comments for sensitive files. Confirmed risks remain because the skill reads an Anthropic API key, sends selected file contents to Claude, can fall back to the Claude CLI, and overwrites the target file with model output after backup.

Files scanned

819

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (2)

High

scripts/compress.py:104-120 scripts/compress.py:192-199

Untrusted Content Sent to External LLM and Written Back

compress.py embeds the original file text in a prompt sent to Claude, then writes the model output over the original path after creating a backup. Sensitive-path denylist checks reduce exposure, but ordinary prose files may still contain private information and prompt-injection text that can alter the compressed result.

The code directly places original file content into an LLM prompt and writes the returned text to disk. The backup and validation reduce data-loss risk, but they do not remove the third-party disclosure or prompt-injection risk.

Medium

SKILL.md:30-31

Network reconnaissance

- call Claude to compress

The documented process calls Claude to compress file contents. This is intended network/API use, not reconnaissance, but it confirms third-party transmission of user-provided file content.

Capability review items (3)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

scripts/compress.py:76

Generic API/secret keys

api_key = os.environ.get("ANTHROPIC_API_KEY")

The line retrieves ANTHROPIC_API_KEY. The key is used for the Anthropic client, so secret handling is real even though it is not logged or exfiltrated.

High

scripts/compress.py:81

Generic API/secret keys

client = anthropic.Anthropic(api_key=api_key)

The API key is passed into the Anthropic client. This is expected authentication but still confirms use of a secret in process memory.

Low

scripts/compress.py:76

Python environment access

api_key = os.environ.get("ANTHROPIC_API_KEY")

The code reads ANTHROPIC_API_KEY from the environment. This is legitimate authentication, but environment secret access is still a real sensitive-data boundary.

Risk Factors

Detected Patterns

Network reconnaissance

Audited by: claude View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/juliusbrussee-compress/audits/6?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/juliusbrussee-compress/security.svg)](https://skillstore.io/skills/juliusbrussee-compress?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/juliusbrussee-compress?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/juliusbrussee-compress/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/juliusbrussee-compress.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

juliusbrussee. (2026). compress security audit report (audit version 6) [Author version unspecified]. Skillstore. https://skillstore.io/skills/juliusbrussee-compress/audits/6

BibTeX citation

@techreport{juliusbrussee-juliusbrussee-compress-2026,
  author = {juliusbrussee},
  title = {compress security audit report (audit version 6)},
  institution = {Skillstore},
  year = {2026},
  number = {6},
  url = {https://skillstore.io/skills/juliusbrussee-compress/audits/6},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "compress security audit report (audit version 6)"
version: "unspecified"
type: report
authors:
  - name: "juliusbrussee"
date-released: "2026-07-09"
url: "https://skillstore.io/skills/juliusbrussee-compress/audits/6"
identifiers:
  - type: other
    value: "skillstore:juliusbrussee-compress:audit:6"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: High

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Reduce project memory size

Compress a long CLAUDE.md file so agents spend fewer tokens reading repeated project instructions.

Shorten planning notes

Turn verbose markdown notes into compact action-oriented text while keeping paths, commands, and examples intact.

Maintain compact team preferences

Compress preference and todo files that are loaded often across AI coding sessions.

Try These Prompts

Compress one memory file

Compress this memory file: <absolute filepath>. Preserve code, links, commands, and headings exactly.

Compress a project instruction file

Use caveman compression on my CLAUDE.md file. Stop if the file is not natural language or if a backup already exists.

Compress notes while preserving structure

Compress this markdown notes file into caveman style. Keep headings, bullets, code blocks, URLs, paths, and commands unchanged.

Validate a compression result

Compress this file and verify the result. If validation fails, restore the original and report the validation errors.

Best Practices

Review the file for private information before compression because contents may be sent to Claude.
Keep the generated backup until you confirm the compressed file is accurate.
Use the skill on markdown or text memory files, not source code or configuration files.