Coupler.io Business Analytics Skills

What does this agent do, and when should you use it?

This is Coupler.io's repository of AI agent skills for marketing and PPC, sales, finance, ecommerce, SEO, GEO and Black Friday analysis. Each skill is a SKILL.md file that spells out a repeatable analysis procedure; the data comes through the Coupler.io MCP server, where tools such as list-skills, get-skill, list-datasets, get-schema and get-data discover datasets, inspect columns and run SQL against live business data. Skills define how to analyze, while the MCP layer handles access and execution, and outputs are turned into a structured, validated report. The repo also ships shared capability skills (get-started, create-dataflow, google-ads-custom-gaql, report-generation) and utilities such as an AI-writing humanizer and a live-dashboard builder. Installation runs through a Claude Code plugin marketplace, per-business plugins, or a manual copy into .claude/skills/, so you can install only the cluster you need.

Skills are organized as folders, each containing a SKILL.md with YAML frontmatter limited to the six portable spec keys (name, description, metadata and so on), carrying short_description, category and sources under metadata; a generator script builds the machine-readable skills-index.json. At runtime a skill calls Coupler.io MCP tools: list-skills and get-skill to load the procedure, list-templates, list-integrations, list-credentials, list-datasets and search-datasets for discovery, get-integration, get-integration-field-options, get-dataflow, get-schema and get-data for inspection, create-dataflow, create-dataflow-source, create-dataflow-destination, update-dataflow-source and update-dataflow-destination for configuration, run-dataflow with preview or full mode, and update-dataset to persist a dataset description. The PPC skill, for example, traces funnels per platform, ranks performance on CPA/ROAS, checks budget pacing and creative fatigue, and deliberately refuses to blend conversions across platforms; google-ads-custom-gaql builds a Custom GAQL source when no packaged report type carries a field. report-generation formats output as TL;DR, Metrics, Context, Recommendations and Next Questions, then runs a validation pass over arithmetic, units and claim-versus-data consistency.

1. A paid-media lead wants to know why Google Ads ROAS fell last week and which campaigns waste budget.
2. An ecommerce operator needs Shopify sales broken down by product, variant, inventory and discount code, plus a stockout forecast.
3. A sales manager reviews pipeline coverage, the weakest stage conversion and rep-level cycle length.
4. A finance team explains this month's P&L movement, calculates cash runway or bridges MRR/ARR changes.
5. An SEO or content team finds queries ranking just off page one, tracks new-page indexing, or diagnoses a sudden organic traffic drop.
6. A growth team decides daily, during Black Friday, which campaigns to scale, cut or leave alone.

How do you install or deploy this agent?

You need a Coupler.io account with the Coupler.io MCP server connected. In Claude Code, add the marketplace and install the plugins you want:

/plugin marketplace add coupler-io/skills
/plugin install coupler-finance@coupler-io-skills
/plugin install coupler-sales@coupler-io-skills
/plugin install coupler-ecommerce@coupler-io-skills
/plugin install coupler-marketing-ads@coupler-io-skills

If the install summary asks you to activate, run:

/reload-plugins

You can inspect a plugin's skill count and cost first:

claude plugin details coupler-marketing-ads@coupler-io-skills

Manual alternative: copy any skill folder containing a SKILL.md into your project's .claude/skills/ directory. For a whole team, register the marketplace in the repository's .claude/settings.json:

{
  "extraKnownMarketplaces": {
    "coupler-io-skills": {
      "source": {
        "source": "github",
        "repo": "coupler-io/skills"
      }
    }
  }
}

This adds the marketplace only; each person still installs plugins with /plugin install.

How do you use this agent?

After installing, ask questions in natural language and Claude will reach for matching skills, or invoke a namespaced skill explicitly:

/coupler-finance:finance-analytics

Questions like "Why did Google Ads ROAS decline last week?" or "Which customer cohorts have the strongest retention?" follow the skill's procedure, starting with SELECT * FROM data LIMIT 5 to confirm the dataset shape. To build a pipeline from zero, use the get-started and create-dataflow skills: choose an integration, attach a credential, wire source to destination, run preview, then run full, and optionally add a refresh schedule. When editing skills, change the SKILL.md frontmatter and regenerate the index:

python .github/scripts/generate_skill_index.py

Validate before submitting:

python .github/scripts/generate_skill_index.py --check

What are this agent's strengths and limitations?

How does this agent compare with similar options?

Key facts side by side with the most closely related agents.

How does FollowAgents rate this agent?

FAQ