An AI skill is a structured system prompt or set of custom instructions that gives AI assistants like ChatGPT, Claude, and Gemini specialized capabilities. It includes a clear role definition, specific knowledge areas, step-by-step processes, and quality criteria that help AI perform specific professional tasks.

How does Skill This work?

Simply describe your professional expertise in plain text. Our free AI skill generator analyzes your input and transforms it into a structured system prompt. The skill is then graded on clarity, specificity, and usefulness. You can share your skill via a unique URL or copy it to use as custom instructions with any AI assistant like ChatGPT, Claude, or Gemini.

Is Skill This free to use?

Yes, Skill This is completely free. No account or payment is required. You can create unlimited skills and share them with anyone.

How is my skill graded?

Skills are graded on a scale from A+ to F (0-100 points). The grade reflects how clear, specific, and actionable your skill is. Higher grades indicate skills that AI assistants can use more effectively.

How do I use my skill with an AI assistant?

Copy your generated skill and paste it into the system prompt or custom instructions of any AI assistant. For ChatGPT, use Personalization > Custom Instructions. For Claude, add it to a Project's instructions. For Cursor or other IDEs, save it as a .cursorrules or CLAUDE.md file. The AI will then adopt that skill and perform tasks according to your expertise.

AI Skill Report Card

Crafting Boolean Search Dorks

A-87·Oct 2, 2026·Source: Extension-selection

Markdown
---
name: crafting-boolean-search-dorks
description: Constructs precise Boolean search strings and Google Dorking queries to bypass AI-generated summaries and retrieve literal, token-matched results from search engines. Use when a user needs to find specific files, exposed data, pre-AI/human-authored content, or wants to force verbatim/non-AI search behavior instead of conversational semantic search.
---

# Crafting Boolean Search Dorks

Quick Start14 / 15

To find PDF reports on a specific domain, published before generative AI content flooded the web, and strip out AI Overviews:

site:example.com filetype:pdf intitle:"annual report" -ai before:2022-11-30

Append &udm=14 to the resulting Google URL to force the classic text-only, link-first layout (no AI Overview, no auto-expansion).

Recommendation▾

Add an example of combining multiple AI-bypass techniques simultaneously to show full-stack suppression in action

Workflow14 / 15

Progress:

Step 1: Clarify the retrieval goal (exact file, exposed data, domain-restricted content, or historical/pre-AI content)
Step 2: Select core Boolean logic (AND / OR / NOT / grouping)
Step 3: Add relevant Dorking operators (site:, filetype:, inurl:, intitle:, intext:, AROUND(N))
Step 4: Apply AI-bypass layer if needed (verbatim mode, -ai, udm=14, tbs=li:1)
Step 5: Apply temporal constraints if isolating pre-AI/human content (before:/after:)
Step 6: Choose platform (Google verbatim, DuckDuckGo No-AI, Mojeek, SymbolHound) based on need
Step 7: Validate query syntax (no spaces around operators, quotes around exact phrases, parentheses balanced)

Step-by-step details

Define intent. Is this OSINT/recon (dorking), a narrow document lookup, or an attempt to avoid synthesized/AI answers? This determines which operator family to use.
Build the Boolean skeleton first.
- AND / & — force intersection (often implicit; use explicit AND for clarity in complex strings)
- OR / | — broaden to synonyms/alternatives — always wrap in parentheses when mixed with AND
- -term — exclude (no space between - and the term)
- () — group sub-expressions to control evaluation order

Layer in Dorking operators (combine freely, space-separated):

Operator	Purpose
`site:`	restrict to domain/TLD
`filetype:` / `ext:`	restrict to file extension
`inurl:`	string must appear in URL path
`intitle:`	string must appear in `<title>`
`intext:`	string must appear in body text
`AROUND(N)`	two terms within N words of each other

Note: related: and cache: are deprecated/removed — do not rely on them.

Force literal/verbatim matching when the engine is auto-correcting, swapping synonyms, or injecting AI summaries:
- Use Google's Verbatim mode (Tools → All Results → Verbatim)
- Append &udm=14 to the URL (text-only web results, no AI Overview)
- Append tbs=li:1 (enforces verbatim)
- Add -ai as an exclusion token to suppress AI/overview content
Apply temporal dorking to isolate pre-generative-AI content:
- before:2022-11-30 — pre-ChatGPT launch cutoff (most precise)
- before:2023 — clean annual cutoff
- before:2019 — strict historical window, pre-transformer-era content
- URL-layer equivalent: &tbs=cdr:1,cd_max=11/29/2022
Switch platforms when Google dorking is throttled or summarized away:
- DuckDuckGo No-AI — strips conversational widgets
- Mojeek — independent index, literal token matching, no semantic synthesis
- SymbolHound — searches special characters/symbols that major engines ignore (useful for code, error messages, syntax)

Recommendation▾

Include a brief note on legal/ethical boundaries of dorking since it touches OSINT/exposed-data use cases

Examples18 / 20

Example 1: Input: "Find exposed admin login pages on .edu domains, excluding AI-related noise." Output:

inurl:admin intitle:"login" site:*.edu -ai

Example 2: Input: "Locate Word resumes mentioning 'machine learning' but not 'internship', published before ChatGPT existed." Output:

filetype:docx intext:"machine learning" -internship before:2022-11-30

Example 3: Input: "Search for pages about 'python OR javascript' tutorials, but not for beginners, and force a non-AI Google results page." Output:

(python OR javascript) tutorial -beginners

Append &udm=14 to the search URL to render classic link-only results.

Example 4: Input: "Find strict pre-2019 forum discussions about 'cold fusion' within 3 words of 'experiment', avoiding modern AI-generated rehashes." Output:

"cold fusion" AROUND(3) experiment before:2019

Run via Mojeek or DuckDuckGo No-AI for additional AI-content filtering.

Recommendation▾

Consider a troubleshooting table mapping symptoms (e.g., 'results still show AI Overview') to specific fixes

Best Practices

Always quote multi-word exact phrases: "index of" not index of.
No space after a prefix operator: filetype:pdf not filetype: pdf; same for -exclude.
Wrap OR clauses in parentheses when combined with AND terms to avoid ambiguous precedence.
Default to before:2022-11-30 when the goal is strictly "pre-AI content" — it's the most defensible, citable cutoff.
Combine -ai + &udm=14 + Verbatim mode together for maximum suppression of synthesized results; one alone is often insufficient.
When a dork chain gets ignored or "softened" by the engine, switch to Mojeek or DuckDuckGo No-AI rather than fighting Google's query reinterpretation.
For symbols/code snippets (e.g., C++, $variable, regex), use SymbolHound instead of mainstream engines, which strip special characters.

Common Pitfalls

Don't rely on related: or cache: — both are deprecated/dead on Google.
Don't assume Boolean operators are absolute on modern engines by default — they're often treated as soft ranking signals unless Verbatim mode or udm=14 is explicitly forced.
Don't put a space between a negative sign and the excluded term (- term fails; -term works).
Don't mix OR and AND without parentheses — results in unpredictable precedence.
Don't expect before:/after: to be pixel-perfect; they filter by Google's inferred publish/crawl date, which can be inaccurate for undated or republished pages.
Don't forget that excluding -ai filters the literal token "ai," which can over-exclude legitimate pages (e.g., about "air" compounds) — test and refine.