AugmentClaude

Compare Options

Compare 2–5 products or plans side-by-side with weighted criteria and a recommendation.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/compare-options-nano-muse/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Safe default: works for every skill.

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Compare two to five options (products, plans, places, tools, offers) on the criteria that matter to the user, in a table the user can re-weight, with a recommendation. Use when the user asks which to pick, for a comparison, or "X vs Y".

What this skill does

Compare options

The result is a page, compare/<topic>.html, and a recommendation in chat with the reason.

Pin down the question

  • The options (2–5). If the user named one and asked for alternatives, find at most three obvious ones with web_search, and say that you picked them.
  • The criteria: what the user said matters, plus recall for standing preferences (budget, brands they avoid, "must work offline", family size…). Five to seven criteria, no more. If the user gave none, propose them in the page; do not ask first.
  • The user's situation in one line at the top of the page (what it is for, budget, constraints), so the recommendation has a visible basis.

Research

  • One web_search per option, then web_fetch the one or two pages that actually contain the facts (the maker's page, a review that lists specifications, a price page). Stop there; more searching rarely changes the answer.
  • Every number in the table comes from a page you read, with the page's host in the cell's tooltip (title attribute). No source, no number: write "n/a" and say so.
  • Prices with the date and currency; anything that varies by region or plan, say so.

The page

  • A table: options as columns, criteria as rows; a weight (1–5) next to each criterion the user can change in the page, and a score line that recomputes live. Plain JavaScript inline; no external resources; readable on a phone (scrolls sideways).
  • Under the table: one paragraph per option: what it is best for and its one real drawback. Then Recommendation: the pick, the reason in two sentences, and what would change the pick ("if you travel a lot, take B").
  • Keep it honest: if two options are close, say they are close.

Finish

Chat reply: the pick, the reason in one sentence, the file name; offer to remember a preference if one came out of it (the user said "I don't care about weight": that is a preference). Never buy, sign up or fill forms; if the user asks you to, that is a new task with its own approval.

Related skills