agentleFS
ZachArticulateVZachArticulateV/designer-pro-and-seo/skills/design-research/SKILL.md

design-research

Skill10 starsChanged 28 days agoMITView on GitHub

Run a competitive research pass on a target site and its top competitors, producing a structured competitive-intelligence report — brand snapshot, competitor scoring, white-space gaps, and keyword landscape — plus a build brief that feeds the design and SEO workflow. Uses the Firecrawl MCP for deep multi-page scraping when connected; otherwise built-in WebFetch/WebSearch + html-extract. Trigger when the user says "competitive research", "competitor analysis", "research the niche", "niche research", "what are competitors doing", "give me a competitive report", "client research", or "engagement research".

  • Reads credentials

10 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Detect tooling: Check whether the Firecrawl MCP is connected:

    line 39
  2. Distinctive:

    Identify competitors: Use WebSearch on the niche + the target's core terms to assemble a competitor set (default 5).

    line 47
  3. Very specific:

    Read the craft dimensions with the negative-space/rhythm method in references/design-research/white-space-method.md.

    line 58
  4. Distinctive:

    Tier 1 — Firecrawl MCP: If connected, deep multi-page scraping of target + competitors.

    line 70
  5. Distinctive:

    Tier 4 — guided: If a site is JS-rendered and Firecrawl isn't connected, work from the pages you can fetch and note what a full crawl would …

    line 74
  6. Distinctive:

    Always state which tier ran and what Firecrawl would add.

    line 77
  7. Distinctive:

    Every row degrades to a real deliverable.

    line 106

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

design-research

Family: design Status: Stable

Purpose

The research front-end of an engagement. Given a target domain and a niche, it captures the target's positioning, identifies top competitors, scores each on key dimensions, surfaces the white space no one is filling, and produces a markdown or interactive-HTML report that feeds design-system-gen, design-dimensions, blast-prompt, and the SEO skills. It is tool-aware: with the Firecrawl MCP it does deep multi-page scraping; without it, built-in web fetch/search + html-extract still produce a real competitive read.

Triggers

  • "competitive research" / "competitor analysis"
  • "research the niche" / "niche research"
  • "what are competitors doing"
  • "give me a competitive report"
  • "client research" / "engagement research"

Inputs

  • Target domain
  • Industry / vertical
  • Geographic scope (local / national / global)
  • Number of competitors (default 5)
  • Output format: markdown | interactive HTML | both

Steps

  1. Detect tooling. Check whether the Firecrawl MCP is connected:
    python3 "${CLAUDE_PLUGIN_ROOT}/scripts/workflow/capability_probe.py"
    (on Windows use py if python3 is absent; in PowerShell the variable is $env:CLAUDE_PLUGIN_ROOT). It reports whether FIRECRAWL_API_KEY is set.
  2. Capture the target. Tier 1: Firecrawl-scrape the site for voice, services, positioning. Tier 2: WebFetch the key pages + html-extract for structure/tokens.
  3. Identify competitors. Use WebSearch on the niche + the target's core terms to assemble a competitor set (default 5); let the user confirm/adjust.
  4. Scan each competitor. Same tier path as step 2, one pass per competitor.
  5. Score each on the 18-dimension rubric in data/competitor-rubric.csv (positioning, brand-craft, content-depth, trust-signals, conversion-path, findability). Score every dimension 0–4 against its good/bad anchors and roll up the weighted composite exactly as references/design-research/competitor-rubric.md describes, so scores compare across competitors.
  6. Find the white space — walk each dimension across the whole set and flag the ones where every competitor is weak (the rubric's white_space_test column). Read the craft dimensions with the negative-space/rhythm method in references/design-research/white-space-method.md.
  7. Keyword landscape — call seo-cluster if available; otherwise a WebSearch-based intent grouping.
  8. Compose the report into the user's workspace (see Outputs).
  9. Report which tier ran and what a full Firecrawl crawl would add.

Capability routing

This skill follows the plugin's capability-tier cascade (references/CAPABILITY-TIERS.md):

  1. Tier 1 — Firecrawl MCP. If connected, deep multi-page scraping of target + competitors.
  2. Tier 2 — built-in (default). WebFetch + WebSearch + html-extract for a real competitive read from the key pages.
  3. Tier 4 — guided. If a site is JS-rendered and Firecrawl isn't connected, work from the pages you can fetch and note what a full crawl would add.

Always state which tier ran and what Firecrawl would add.

Outputs

Written into the user's project workspace:

OutputWhat it containsFormatQuality bar (how it is scored)
research/01-target-brand.mdbrand/positioning extraction for the targetmdevery claim traces to a fetched page; no invented positioning
research/02-competitor-analysis.mdeach competitor scored 0–4 on all 18 rubric dimensions + six weighted category scores + overall + keyword landscapemd (Critical/High/Medium open-lane groups)every score cites the signal_to_check it was read from; anchors from data/competitor-rubric.csv; unobservable dimensions marked "not assessed", never zeroed
research/03-build-brief.mdthe highest-weight open lanes turned into headline build movesmdeach move names the dimension + white_space_test it fills; ranked by rubric weight
competitive-analysis.html (optional)interactive scoring report (radar/bar over the six categories)HTMLscale reads 0–4 honestly; no rescale-to-100 implying false precision
Tier linewhich tier ran + what a full Firecrawl crawl would addone sentencestates the tier honestly; no fabricated SEO/performance number

Filed to: the user's project workspace, never into the plugin. A never-fabricate field (competitor SEO volume, field CWV, backlink counts) ships as a labeled proxy

  • a needs_tier1 note — never a synthesized number.

Error Handling

ConditionDetectionBehavior (degrade, never fail)User-facing message
Firecrawl MCP absentcapability_probe.py reports FIRECRAWL_API_KEY unset / MCP not exposedfall through to Tier-2 WebFetch + WebSearch + html-extract"Ran Tier 2 (built-in fetch). Add Firecrawl for deep multi-page + JS-rendered crawl."
A competitor site is JS-gated / auth-walledkey pages fetch empty or a shellscore only the dimensions whose signal_to_check was observable; mark the rest "not assessed""Couldn't reach <pages> — scored the visible signals; a Tier-1 crawl would complete the rubric."
No network / offlinefetch raises or --no-network setanalyze provided saved pages / files only"Offline — scored the provided pages; live fetch would add competitors and freshness signals."
Too few competitors foundWebSearch returns < requested setscore what was found; state the smaller n"Found only <n> credible competitors — scored those; broaden the niche terms to add more."
seo-cluster unavailableskill not installed / not routedfall back to a WebSearch-based intent grouping for the keyword landscape"Used a WebSearch keyword read; install seo-cluster for SERP-overlap clustering."
Bad / empty input (no target domain)required input missingask for the target + niche; do not invent a subject"A target domain and niche are required to run a competitive pass."

Every row degrades to a real deliverable; the built-in Tier-2 path is the product.

Dependencies

  • Optional (Tier 1): Firecrawl MCP (extensions/firecrawl/) — deep scraping; free path works without it
  • Built-in (Tier 2): WebFetch; WebSearch; html-extract
  • seo-cluster (optional — adds SERP-based keyword landscape; free path: WebSearch intent grouping)
  • scripts/workflow/capability_probe.py (required — tooling detection)
  • Scoring rubric + method (required knowledge): data/competitor-rubric.csv, references/design-research/competitor-rubric.md, references/design-research/white-space-method.md

Notes

The output structure is a proven engagement shape, made repeatable here instead of bespoke per project. Respect target sites' robots.txt and Terms of Service when scraping (see PRIVACY.md); research artifacts stay in the user's workspace.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.