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

Building Personal Search Indexes

B68·Nov 6, 2026·Source: Extension-page

Quick Start13 / 15

Minimal setup flow for a local, privacy-first search engine:

Bash
# 1. Install/run the indexer (example pattern)
docker run -d --name hister \
  -p 8080:8080 \
  -v hister_data:/data \
  hister/hister:latest

# 2. Point it at sources to index
hister crawl add --type browser-history --since 30d
hister crawl add --type local-dir --path ~/Documents --recursive
hister crawl add --type url --seed https://example.com --depth 2

# 3. Run the crawl/index job
hister index run

# 4. Query it
hister search 'title:"project plan" AND modified>2024-01-01'

If no tool is specified, default to Hister's model: local daemon + crawler + full-text index + query language, with AI/semantic search as an optional add-on layer, not a requirement.

Recommendation▾

The skill is built almost entirely around a single, seemingly fictional/niche tool ('Hister') without clarifying whether it's real or hypothetical—this risks hallucinated command syntax being presented as fact; verify tool existence or generalize to real tools (e.g., Recoll, DocFetcher, Meilisearch).

Workflow13 / 15

Progress:

Step 1: Define sources (browser history, local directories, seed URLs)
Step 2: Configure privacy boundaries (what stays local, what's excluded)
Step 3: Set crawler mode (headless browser vs. lightweight HTTP fetcher)
Step 4: Run initial full index
Step 5: Set up incremental/scheduled re-indexing
Step 6: Design queries using the advanced query language
Step 7 (optional): Enable AI/semantic layer for fuzzy or conceptual search
Step 8 (optional): Expose as multi-user/team instance with access controls

Step details

1. Define sources

Browser history: export or live-sync from browser profile dirs.
Local files: pick directories with real content (docs, notes, PDFs), exclude caches/binaries.
Web: seed URLs + crawl depth for sites you want fully indexed (wikis, docs sites).

2. Privacy boundaries

Keep index and raw content on local disk or private network only.
Exclude sensitive paths (password managers, .env, private keys) explicitly before first crawl.
For multi-user deployments, separate per-user indexes unless a shared team knowledge base is intended.

3. Crawler mode

Use a real browser engine (headless Chrome/Playwright-style) for JS-heavy sites or when capturing rendered page text.
Use a traditional HTTP fetcher for speed on static sites/large local file sets.

4-5. Indexing

Full index once; switch to incremental (diff-based on mtime/ETag) for ongoing updates.
Schedule via cron or built-in scheduler (e.g., nightly for browsing history, weekly for static doc sets).

6. Query language

Favor structured queries over keyword soup: field filters (title:, url:, path:), boolean operators (AND/OR/NOT), date ranges, phrase matching with quotes.
Combine full-text terms with metadata filters to cut noise fast.

7. AI/semantic layer (optional)

Layer embeddings-based search on top of, not instead of, full-text index — use it for "find similar/related" and conceptual recall, keep exact-match query language for precision tasks.
If MCP integration is available, expose search as a tool so AI assistants can query the local index directly.

8. Multi-user/team

Add auth and per-user index scoping before exposing beyond localhost.
Decide shared vs. isolated indexes per team/project.

Recommendation▾

Add a troubleshooting section for common setup failures (crawler fails on JS sites, index corruption, permission errors) rather than just conceptual pitfalls.

Examples14 / 20

Example 1: Input: "I want to search everything I've read on arxiv.org in the last year, plus my local PDF notes." Output:

hister crawl add --type url --seed https://arxiv.org --filter "path:/abs/*" --since 365d
hister crawl add --type local-dir --path ~/notes --filetypes pdf,md
hister search 'domain:arxiv.org AND after:2023-01-01 AND "transformer"'

Example 2: Input: "Set up a private team search engine for our internal wiki and shared drive." Output:

hister crawl add --type url --seed https://wiki.internal --depth 5 --auth cookie.json
hister crawl add --type network-share --path //fileserver/team-docs
hister instance enable-multiuser --auth oidc
hister search 'path:"team-docs/Q3" AND title:roadmap'

Example 3: Input: "Find that article about rate limiting I read a few weeks ago but can't remember the exact title." Output:

hister search 'rate limiting AND visited:14d..30d'
# If exact terms fail, fall back to semantic search:
hister search --semantic "article about throttling API requests, read recently"

Recommendation▾

Examples all use the same invented CLI syntax—show a bad/failed query example (e.g., malformed query syntax or an overly broad crawl) alongside the corrected version to meet 'good and bad outcomes' criteria.

Best Practices

Default to full-text + query language first; reach for semantic/AI search only when exact queries fail.
Re-index incrementally, not from scratch, to keep crawl times low and avoid redundant storage.
Exclude noisy/low-value sources (ad trackers, redirect pages, binary caches) from browsing history indexing.
Keep raw crawled content and the index itself on encrypted local storage if the machine is shared or portable.
Use depth limits on web crawls — unbounded crawls balloon storage and index junk.
Version or snapshot the index before major re-crawls so you can roll back.

Common Pitfalls

Indexing entire browser history unfiltered — it captures logins, banking, and other sensitive pages; always exclude/allowlist.
Running a headless-browser crawler on thousands of static pages — wastes resources; reserve it for JS-rendered content only.
Treating semantic search as a replacement for structured queries — it's fuzzier and slower; use field-based queries when you know what you're looking for.
Exposing a local instance to the network without auth "just for a quick team test" — lock down access before binding beyond localhost.
Letting local file indexing recurse into node_modules, .git, or build artifacts — exclude these explicitly to avoid index bloat.