competitor-claims-plugin
Ryanthelion09/competitor-claims-plugin
Claude Code plugin: re-runnable competitive analysis of AI cybersecurity product claims (multi-agent live website sweeps, adversarial verification, dated PDF reports)
Instalação
npx convez-ai Ryanthelion09/competitor-claims-plugin1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
competitor-claims
A Claude Code plugin that runs a fresh, comprehensive comparison of the product claims made by AI cybersecurity companies (Kai, Armadin, Astelia, Artemis, Cogent, Vega, Zafran) every time it's invoked — because marketing claims drift over time.
What it does
Each run:
- Gathers live data via a multi-agent workflow: one explore agent per company (8–15 pages each: product, use cases, integrations, customers, pricing, trust/compliance, blog), an adversarial fact-checker per company that re-fetches cited pages to catch misquotes and overstatements, and three third-party sweeps (buyer reviews, news/funding/analyst coverage, head-to-head comparisons). Sites that block crawlers fall back to the Internet Archive (snapshot date recorded).
- Synthesizes a report focused on how each company markets its product: executive summary, dimension-by-company comparison table, claimed use-case matrix, per-company profiles (taglines, claimed use cases under the vendor's own names, marketing themes, verbatim-quoted claims with source URLs, verification flags), third-party perspective, cross-cutting observations, and a "changes since last run" diff against the previous report.
- Asks before saving the report as a dated PDF (plus markdown source) in a
Generated Analysis/subfolder of the current project.
Installation
/plugin marketplace add https://github.com/Ryanthelion09/competitor-claims-plugin
/plugin install competitor-claims@kai-tools
Usage
/competitor-claims
Optionally pass extra companies or a focus area as arguments. To change the standing roster, edit the table in skills/competitor-claims/SKILL.md.
Requirements
- Claude Code with web access (WebFetch/WebSearch) and the Workflow tool — runs spawn ~15 subagents and take 10–20 minutes.
python3(markdown→HTML rendering; stdlib only).- An HTML→PDF converter — tries Chrome/Chromium/Edge/Brave headless, then
wkhtmltopdf,pandoc,weasyprint. Without one, the report is saved as HTML+markdown instead.
Files
| File | Purpose |
|---|---|
skills/competitor-claims/SKILL.md | The repeatable procedure (roster, workflow invocation, report structure, save flow) |
skills/competitor-claims/claims-workflow.js | Multi-agent orchestration script (explore → verify → third-party) |
skills/competitor-claims/md-to-html.py | Print-styled markdown→HTML converter for the PDF |
skills/competitor-claims/html-to-pdf.sh | HTML→PDF with converter fallback chain |