Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
<p align="center"> <img src="./assets/banner.png" alt="CompetLab Agent Skills — Competitive Intelligence for AI Agents" width="100%" /> </p>
CompetLab Agent Skills
Competitive intelligence your agent can read — and read correctly.
CompetLab monitors your competitors across six dimensions and writes a Strategic Briefing over fourteen analysis areas. These nine skills give an AI agent the vocabulary to read that data the way the platform intends, and to shape it into the thing you actually need — a pulse, a dossier, a battlecard, a work list.
The skills do not do the research. The platform does, with more sources and a memory of prior readings. The skills make sure what comes out the other end is true.
Install
git clone https://github.com/competlab/competlab-ci-skills.git
mkdir -p .claude/skills
cp -r competlab-ci-skills/skills/. .claude/skills/Or, in Claude Code:
/plugin marketplace add competlab/competlab-ci-skills
/plugin install competlab-ci-skills@competlab-ci-skillsOr with the skills CLI:
npx skills add competlab/competlab-ci-skills --all # everything
npx skills add competlab/competlab-ci-skills --skill competlab-ai-visibility # one
npx skills add competlab/competlab-ci-skills --all -g # all projectsAll three work. Every skill carries what it needs inside its own folder, so nothing depends on which path you chose.
Then connect the MCP server, below. Every skill needs it — competlab-site-audit is the one that needs nothing else.
Connect CompetLab
claude mcp add --transport http competlab https://mcp.competlab.com/mcp \
--header "CL-API-Key: cl_live_your_key_here"<details> <summary>Other clients</summary>
{
"mcpServers": {
"competlab": {
"type": "http",
"url": "https://mcp.competlab.com/mcp",
"headers": { "CL-API-Key": "cl_live_your_key_here" }
}
}
}</details>
Keys come from app.competlab.com → Organization Settings → API Keys. One key covers your whole organization, and a read-only key is enough for everything here. Setup guide: competlab.com/developers/mcp.
Start a free trial — 14 days, no credit card. Your API key is available immediately.
The skills
| Skill | What it answers | Say this |
|---|---|---|
| [competlab-ai-visibility](skills/competlab-ai-visibility/) | Which companies do AI models recommend in my category — and am I one of them? | "Are we in the core?" |
| [competlab-ai-sources](skills/competlab-ai-sources/) | Which pages do the engines read before answering, and who is named on them? | "Why them and not us?" |
| [competlab-pricing](skills/competlab-pricing/) | What does each company charge, what would a typical buyer pay at each, and what changed since the last check? | "Did anyone change their prices?" |
| [competlab-positioning](skills/competlab-positioning/) | What does each company say on its homepage, what do I say that nobody else does, what changed, and what do pages say about me? | "Did anyone change their headline?" |
| [competlab-briefing](skills/competlab-briefing/) | What changed, what it means, what to do — at pulse, landscape or dimension depth | "Catch me up" |
| [competlab-competitor-dive](skills/competlab-competitor-dive/) | Everything we know about one rival, across six dimensions | "Deep dive on [competitor]" |
| [competlab-battlecard](skills/competlab-battlecard/) | A 60-second sales reference for a live call | "Battlecard vs [competitor]" |
| [competlab-site-audit](skills/competlab-site-audit/) | What can a machine actually read on this site? No project needed | "Audit example.com" |
| [competlab-monitoring-setup](skills/competlab-monitoring-setup/) | Am I watching the right competitors and asking the right questions? | "Is my monitoring set up right?" |
Try it on any domain
competlab-site-audit runs on any public domain — no project, no monitored competitors, nothing set up. A free-trial key is enough. Crawler access, sitemap coverage, agent adoption, tech stack, trust signals.
"Audit example.com for agent adoption"
What CompetLab monitors
| Dimension | What it tracks |
|---|---|
| AI Visibility | Which companies AI models recommend in your category, and where you sit among them |
| AI Sources | The pages Perplexity and Google AI Overviews read while answering your buyers' questions |
| Positioning | What each company says on its homepage as the page says it, what you say that nobody else does, what changed, and what pages say about you |
| Pricing Intelligence | What each company charges on its pricing page as a buyer in the US sees it, what typical buyer setups cost cheapest first, and every change from X to Y with its dates |
| Content Intelligence | Sitemap analysis, categories, publishing activity, gaps |
| Tech & Trust Profile | Tech stack, security headers, trust signals, AI crawler access |
AI Visibility is queried across five engines — ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews. It answers one question: who is recommended in this category, and are you one of them. Google AI Overviews names companies in prose and ranks nothing.
AI Sources is its companion. It opens up the next question — why them and not you — by showing what the engines actually read. It does not hand you a checklist, because the factors differ by category: uptime-monitoring leaders are weighed on distributed networks, pricing and reliability; database tools on certifications and an entirely different set.
The part that matters most
Every skill carries references/reading-the-data.md, which is the reporting discipline the platform enforces on itself:
- `null` means we did not measure it. Never zero, never empty, never "no". A real
0is a finding;
a null is a gap in our reading, and the two must never share a sentence. One carve-out: a rank is null on a brand named in no answer, and that is measured — not named in any answer.
- A page is not the whole list. The long lists arrive one page at a time; the count is the page
object's total, never the rows you hold.
- Counts, never rates. "Named in 8 of 69 answers", not "12%". The question set is small by design
and a share computed from it is false precision.
- Per engine, never pooled. The engines read different pages; a combined figure describes a list
none of them produced.
- Retrieved, never cited. The engines do not disclose which pages they leaned on, so a citation
count is a number that does not exist.
- Ranges that overlap are not ordered. Two brands whose confidence intervals overlap are tied, and
a difference between two overlapping readings is two readings — not a movement.
- A price is the page's number. Never converted, summed or averaged across companies; a change is
written from X to Y with its two dates, never with an arrow.
An agent given competitive data will happily round, pool and rank it into something confident and wrong. This is the file that stops it.
Cross-agent compatibility
Built on the Agent Skills standard — plain SKILL.md files that work with Claude Code, Cursor, Codex, Gemini CLI and any agent that reads the format. Tool names are written as Claude Code's mcp__competlab__*; other clients may need them adjusted to their own convention.
Contributing
See CONTRIBUTING.md. CI checks that every skill's allowed-tools matches the tools its body invokes, that every referenced file exists, and that the facts about the product have not gone stale. Run it locally:
npm run checkLicense
MIT — see LICENSE.
<p align="center"> <a href="https://competlab.com">CompetLab</a> · <a href="https://competlab.com/developers/mcp">MCP</a> · <a href="https://competlab.com/developers/api">API</a> · <a href="https://competlab.com/developers/sdk">SDK</a> </p>