undetectable-ai
3 findings (1 high, 2 medium), mainly social engineering. Most recent scan 2026-02-07.
This skill poses significant ethical and policy concerns rather than traditional security threats. Its explicit purpose is to help users evade AI detection systems to pass off AI-generated content as human-written, directly facilitating academic dishonesty, plagiarism, and deceptive content submission. While the skill does not contain traditional security vulnerabilities (no data exfiltration, command injection, or credential theft), it is designed to enable systematic deception at scale. The technical implementation appears straightforward (TypeScript CLI tools with pattern matching and text transformation), but the use case is fundamentally problematic. The skill should be rejected based on its intended purpose of facilitating fraud and policy violations, even though it lacks conventional security exploits.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Make AI text undetectable. TypeScript CLI that scans ChatGPT, Claude, OpenClaw output for patterns flagged by GPTZero, Turnitin, Originality.ai, then auto-fixes them to pass as human-written. Use to bypass AI detection, evade checkers, or humanize text before submission.
- Package
- 6 files, 21 KB
- SKILL.md
- 320 words, 12+ sections
- Code
- TypeScript (2)
- Allowed tools
ReadWriteStrReplaceShellGlob
SKILL.md covers: Undetectable AI · Setup · Scripts · analyze.ts — Scan for Detectable Patterns · Analyze a file · From stdin · JSON output · transform.ts — Auto-Fix Patterns · Transform and print · Write to file · Fix em dashes too · Quiet mode
Findings
HIGH Skill Designed to Facilitate Academic Dishonesty and Deception Social Engineeringllm
This skill's explicit purpose is to help users evade AI detection systems (GPTZero, Turnitin, Originality.ai) by making AI-generated content appear human-written. The description states 'bypass AI detection, evade checkers, or humanize text before submission' and the workflow includes scanning, auto-fixing, and re-scanning to 'verify' evasion success. This directly facilitates academic dishonesty, plagiarism, and deceptive content submission in educational and professional contexts.
SKILL.md
Description: 'Make AI text undetectable...auto-fixes them to pass as human-written. Use to bypass AI detection, evade checkers, or humanize text before submission.' Workflow step 4: 'Re-scan to verify' evasion success.MEDIUM Deceptive Framing as Legitimate Writing Tool Social Engineeringllm
The skill frames AI detection evasion as a neutral technical capability ('TypeScript CLI tools', 'removing telltale patterns') without acknowledging the ethical implications or primary use case (academic dishonesty). The clinical, technical presentation obscures that the tool's main purpose is to help users deceive detection systems and misrepresent AI-generated work as their own.
SKILL.md
Title: 'Undetectable AI' with subtitle 'TypeScript CLI tools for bypassing AI content detectors' presented as neutral technical documentation without ethical warnings or legitimate use case justification.MEDIUM Facilitates Systematic Deception at Scale Social Engineeringllm
The skill includes batch processing capabilities ('Scan all docs', 'Transform all') that enable users to systematically process multiple documents to evade detection. This industrializes the deception process, making it easier to submit large volumes of AI-generated content fraudulently across multiple assignments or submissions.
SKILL.md
Batch processing examples: 'for f in *.txt; do...npx ts-node scripts/analyze.ts "$f"...done' and 'for f in *.md; do...npx ts-node scripts/transform.ts "$f"...' enabling mass processing of documents.What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 2.7 KB |
package-lock.json | config | 8.4 KB |
package.json | config | 306 B |
scripts/analyze.ts | script | 3.6 KB |
scripts/patterns.json | config | 1.8 KB |
scripts/transform.ts | script | 4.0 KB |
External hosts referenced
registry.npmjs.org
Hostnames that appear in the skill's files. Referencing a host is not a finding; compare against the findings above and the skill's stated purpose.
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/undetectable-ai
- Raw scan result: undetectable-ai-scan.json
- Badge for your README:
- Request a re-scan