FIND YOUR NEXT THING TO TRY

#claude

A small corner of the library, connected by one useful idea.

6 things to try

🧰
OFP SCORE9.4/10
Work● WORKS

Build My Claude Skill

Turn a task you repeat into a reusable Claude Skill—with guardrails, examples and tests included.

WHY 9.4?

It turns a repeated workflow into a compact Skill with precise triggers, explicit permission boundaries, complete files and three pass/fail evaluations. Not 10/10: generated Skills still need human review and real testing in the target Claude environment.

🧭
OFP SCORE9.3/10
Work● WORKS

Find the Jobs Your CV Actually Fits

Turn one CV into realistic adjacent roles—with evidence, gaps and a practical search plan.

WHY 9.3?

It ties role suggestions to evidence in the CV, separates direct from stretch fits and refuses to invent salary or demand without live sources. Not 10/10: the output still depends on CV quality, access to current listings and human judgment about culture and motivation.

🛠️
OFP SCORE9.5/10
Learn● WORKS

Make My Prompt Better First

Turn a vague request into a prompt worth running—before the AI answers it.

WHY 9.5?

It fixes ambiguity before generation, asks only decision-changing questions and lets you approve the rewritten prompt before spending time on the answer.

🧳
OFP SCORE9.4/10
Work● WORKS

Move This Chat to Another AI

Create a clean context passport and continue in ChatGPT, Claude or Gemini.

WHY 9.4?

The compact handoff preserves decisions and open questions while removing filler, unsupported claims and sensitive details.

⌨️
OFP SCORE8.9/10
Work● WORKS

Give This Chat Four Useful Shortcuts

Teach one chat four tiny commands for pushback, brevity, options and checking.

WHY 8.9?

The commands are easy to remember and genuinely speed up iteration, though they apply only where the AI follows the current chat context.

⌚
OFP SCORE9.1/10
Research● WORKS

Coach Me From My Garmin Data

Turn a Garmin export into a cautious, evidence-based training review.

WHY 9.1?

It checks the available fields, separates evidence from inference and builds a conservative plan without hiding the paid connector requirement.