Who this is for
Developers, DevOps engineers, and solo founders using Claude Code who want Claude to do more than generate boilerplate. Skills are useful when you have a specific workflow that Claude does well once you've told it what to do — but you're tired of telling it every session.
These six skills cover the work that's unglamorous but high-value: parsing structured data out of ugly sources, catching bugs before they ship, auditing the environment before a deploy, and generating infrastructure you can actually read out loud.
Parsing and Debug
PDF Whisperer
Category: Parsing · View skill
PDFs are the worst input format in software. Invoices, contracts, bank statements, government forms, scanned documents — they all arrive as PDFs, and they all need to become structured data before you can do anything with them.
PDF Whisperer gives Claude a systematic extraction workflow: identify the document type, locate the key fields, handle multi-page layouts, deal with tables that span pages, and output clean JSON. The skill's instructions handle the patterns that break naive extraction — merged cells, rotated text headers, footnotes that contain data, and multiple document variants in a single file.
Trigger phrases:
- "Extract the line items from this invoice as JSON."
- "Parse this contract and give me all the dates, parties, and payment terms."
- "Turn this bank statement into a CSV-ready structure."
Regex Therapist
Category: Debug · View skill
Regex breaks in three ways: it matches things it shouldn't, it misses things it should catch, and it works in one regex flavor but not another (Python vs. JavaScript vs. PCRE are not the same). The Regex Therapist skill pre-loads Claude with a debug protocol: analyze the pattern, explain what it actually matches step by step, identify edge cases that break it, check for flavor compatibility, and suggest a corrected version with a test set.
Trigger phrases:
- "Debug this regex — it's not matching what I expect:
[pattern]" - "Explain what this pattern actually matches, including edge cases."
- "This regex works in Python but fails in JavaScript. Why?"
Data and Backend
Schema Sketcher
Category: Database · View skill
The hardest part of designing a database schema isn't knowing SQL — it's translating a fuzzy product requirement into a concrete set of tables that won't fall apart in six months. Schema Sketcher encodes a structured approach to that translation: identify entities and relationships, define columns with appropriate Postgres types, add standard audit columns, write CREATE TABLE statements, and generate a migration file.
The skill generates schemas you can read out loud to a non-technical stakeholder and have them confirm you got it right. It also flags normalization tradeoffs and suggests indexes without being asked.
Trigger phrases:
- "I need to store multi-tenant user data with organizations, workspaces, and members. Give me a Postgres schema."
- "Convert this data model to a database schema with migrations."
- "Design the tables for a simple CRM: contacts, companies, deals."
Testing
Test Goblin
Category: Testing · View skill
Claude can write tests. What it won't do without instruction is go looking for the tests that are actually hard to write — the ones that expose real bugs. Test Goblin changes that. The skill tells Claude to analyze the code for untested branches, and specifically to write tests for the uncomfortable cases: empty arrays where the code assumes non-empty, off-by-one errors at boundaries, Unicode input where ASCII was assumed, time-zone-dependent logic, and leap day edge cases.
These are the tests developers skip because they're annoying to write and rarely catch bugs in normal usage. They catch bugs in production at 2am instead.
Trigger phrases:
- "Find the untested branches in this function and write tests for the edge cases."
- "What are the worst-case inputs for this code? Write tests for them."
- "Review this function for test gaps — especially boundary conditions."
DevOps and Environment
Env Doctor
Category: DevOps · View skill
Missing environment variables don't fail at startup — they fail at the worst possible moment during a deploy, or when a specific code path is hit for the first time in production. Env Doctor gives Claude a pre-deploy audit workflow: list all required variables, check which ones are set, flag missing or empty values, catch duplicated keys, and identify variables that exist but have obviously wrong formats (e.g., a database URL that's missing the port).
It's the check you'd run anyway, encoded so you don't have to remember to run it.
Trigger phrases:
- "Audit my environment variables before the deploy."
- "Check my .env file for missing or malformed values."
- "Something's wrong with my environment setup — can you check it?"
Cron Translator
Category: DevOps · View skill
Cron syntax is not intuitive. "Run at 9am every weekday" is not 0 9 * * 1-5 to most people — it has to be looked up, and then verified, and then explained to whoever reviews the PR. Cron Translator handles the translation in both directions: natural language to cron expression, and cron expression to plain English. It also includes a 30-day calendar preview of when the expression fires, and — importantly — flags schedules that cron can't express (like "every two weeks starting on a specific date") so you know when you need a different solution.
Trigger phrases:
- "Convert this to a cron expression: run every weekday at 9am Eastern."
- "What does this cron expression mean:
0 */6 * * *?" - "Generate a cron expression for a job that runs on the 1st and 15th of every month."
Developer workflow: stacking these skills
These skills compound. A typical feature development sequence might look like:
- Env Doctor before starting — confirm the environment is clean
- Schema Sketcher for the data model — get the Postgres schema right before writing any code
- Test Goblin on the finished feature — catch the edge cases before the PR
- Env Doctor again before deploy — catch any new variables the feature introduced
PDF Whisperer, Regex Therapist, and Cron Translator are utilities — reach for them when the specific task comes up, rather than as part of a fixed workflow.
The developer sweet spot
The skills that work best for developers are the ones that encode "annoying but critical" checks. Not the glamorous parts of development — the audits, the edge-case tests, the infrastructure boilerplate, the parsing tasks that are tedious to get right. Those are exactly the tasks that get skipped when you're moving fast, and exactly the ones that come back as production incidents.
Skills are most useful when the task is well-defined, repeatable, and has a clear expected output. That describes most of the work on this list.