From pm-discovery
Run a competitive benchmark and extract Gold Nuggets for the Steal phase of discovery. Use when the user wants to analyze competitors, says "fais un benchmark concurrentiel", "analyse les solutions existantes", "steal phase", or "trouve des gold nuggets".
How this skill is triggered — by the user, by Claude, or both
Slash command
/pm-discovery:discovery-competitive-benchmarkThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill when the user wants to run a competitive benchmark during a discovery — e.g. "fais un benchmark concurrentiel", "analyse les solutions existantes", "steal phase", "what are competitors doing on X", "trouve des gold nuggets pour ma discovery".
Use this skill when the user wants to run a competitive benchmark during a discovery — e.g. "fais un benchmark concurrentiel", "analyse les solutions existantes", "steal phase", "what are competitors doing on X", "trouve des gold nuggets pour ma discovery".
At Mayday, the Steal phase (step 5 of FOCUSED) is dedicated to benchmarking existing solutions before designing anything. The goal is NOT to copy, but to identify patterns, UX models, and "Gold Nuggets" — the 5 most inspiring solutions that can inform Mayday's design.
Benchmark sources to cover:
Pro tip: Use web search to find changelogs, product documentation, and reviews — these are more accurate than marketing pages.
Ask the user for the following if not already provided:
For each product analyzed, produce a structured entry:
[Product Name] — [One-line description]
Analyze at minimum:
After analyzing all products, extract the 5 most inspiring patterns or solutions — things that should influence Mayday's design:
🥇 Gold Nugget 1: [Short title]
[What it is, why it's powerful, what Mayday should steal from it]
Repeat for 5 nuggets total.
End with a brief synthesis section:
🥇 Gold Nugget: Cursor — Multi-file diff review
When an AI agent modifies multiple files, Cursor shows each impacted file in a sidebar with a before/after diff, letting the user accept or reject changes file by file. The key insight: each change includes a brief rationale ("why this file was touched"), which builds trust. For Mayday: this "impacted document list + rationale + per-item accept/reject" is the exact mental model for KB quality recommendations.
npx claudepluginhub mathieu-cozian/mathieu-pm-toolkit --plugin pm-discoveryCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.