# clamp-sh/mcp [Health: Active]
**Category:** π Monitoring
**Repository:** https://github.com/clamp-sh/mcp
**GitHub Stars:** 2
**npm Downloads (last month):** 284
**Views:** 2
**Installs:** 0
**Upvotes:** 0
**Directory Page:** https://allmcps.com/mcp/clamp-sh-mcp
## Description
AI-native web analytics. Query pageviews, top pages, referrers, countries, devices, and custom events. Create conversion funnels and alerts.
## Tools
Capabilities this server exposes over MCP:
- **traffic.overview** β Pageviews, visitors, sessions, bounce rate, avg duration. Period comparison included.
- **traffic.timeseries** β Any core metric over time with automatic granularity (visitors, sessions, bounce rate, avg duration, or event counts).
- **traffic.compare** β Compare any metric across two arbitrary periods. Returns absolute and percent delta.
- **traffic.live** β Visitors active in the last N minutes plus their top pages, referrers, and countries.
- **breakdown** β Group visitors and pageviews by any dimension: `pathname`, `referrer_host`, `country`, `city`, `region`, `device_type`, `browser`, `browser_version`, `os`, `os_version`, `entry_page`, `exit_page`, `channel`, or any UTM field.
- **pages.engagement** β Per-page metrics with `view`: `summary`, `engagement` (adds engagement seconds and bounce rate), `sections` (per-section view counts for one pathname; needs the section-views SDK extension).
- **events.list** β Custom event counts with property filtering and grouping.
- **events.observed_schema** β The actual fired-event signature with per-property type observations. Diff against a local `event-schema.yaml` to surface schema drift.
- **events.property_values** β Top distinct string-typed values a property has taken on a specific event. Use to discover the value space before defining a cohort filter or running `events.list` with `property=`.
- **revenue.sum** β Sum Money-typed event properties. Split by currency, group by any dimension, switch to first-touch via `attribution_model`. Drill-in filters: `plan`, `product`, `first_touch_dim` + `first_touch_value`.
- **revenue.summary** β The whole Revenue tab in one call: revenue per currency, customers, orders, AOV, LTV, MRR (running balance) + ARR, per-plan rollup (where `mrr_contribution` is the cumulative MRR balance per plan, not the in-period delta), per-product rollup, prior-period comparison, and mode flags (`is_subscriptioβ¦
- **revenue.retention** β Cohort revenue retention on MRR-on-MRR math. Per window: `mature_size`, `retained`, `retention_rate`, `mrr_at_d`, `baseline_mrr_mature`, `nrr` (net revenue retention; >1.0 = net expansion). Anchored at the customer's first `subscription_started`.
- **revenue.timeseries** β Plot one revenue metric per bucket. `metric` accepts `revenue` (Money sum), `mrr` (running balance, picks up from prior period), `customers` (distinct paying), or `transactions` (event count).
- **sessions.paths** β Aggregate session paths: top entry β exit pairs with pages per session and duration.
- **users.journey** β Chronological session-and-event reconstruction for one anonymous ID.
- **cohorts.create** β Define cohorts by event + period + filter; query retention curves; compare 2β10 cohorts side-by-side on the same retention windows; delete a saved cohort by name (definitions are immutable, so delete + re-create is the rename/restructure flow).
- **errors.list** β Recent errors, fingerprint-grouped errors with affected-user counts, error rate over time, and breadcrumbs leading to a single error.
- **projects.list** β List all projects this credential can access.
- **docs.search** β Keyword-search the Clamp docs index.
- **funnels.create** β Define and immediately evaluate a multi-step conversion funnel. Steps accept property predicates: `cta_click[location=hero][plan=pro]`. Names must be unique within a project β a duplicate-name call returns 409.
- **funnels.list** β List funnels or fetch one with cohort filters (country, channel, device, UTM).
- **funnels.delete** β Delete a saved funnel by name (definitions are immutable, so delete + re-create is the rename/restructure flow).
- **alerts.create** β Set up metric alerts (e.g. "visitors drops_by 30% over 7d").
- **alerts.list** β List and remove alerts.
## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):
```json
"mcpServers": {
"mcp": {
"command": "npx",
"args": ["-y","@clamp-sh/mcp"],
"env": {
"CLAMP_API_KEY": ""
}
}
}
```
**Requires environment variables:** `CLAMP_API_KEY` β the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).
## Documentation & README
# @clamp-sh/mcp
Analytics [MCP](https://modelcontextprotocol.io) server for [Clamp Analytics](https://clamp.sh). Gives Claude, Cursor, VS Code Copilot, Windsurf, Cline, and any other Model Context Protocol client live access to your traffic, pages, events, revenue, errors, funnels, cohorts, and alerts β and the tools to create funnels and metric alerts from the conversation.
> **Using Cursor, VS Code, Claude Code, or Claude Desktop?** Install the hosted Remote MCP from your [Clamp dashboard](https://clamp.sh/dashboard) in one click β auth happens automatically, no API key to paste, no config files to edit. Open a project, go to the Installation tab, and pick your editor. This npm package is the stdio path for CI, headless agents, custom clients, and self-hosted Clamp instances.
## What your agent can do
Once connected, your agent can answer questions like:
- "How's traffic this week?"
- "Why did pageviews drop on Tuesday?"
- "Which referrers drove the most signups?"
- "Compare US vs Germany conversion this month"
- "Create a funnel from /pricing to signup to purchase"
- "Alert me when bounce rate on /pricing goes above 70%"
- "What did Hacker News visitors actually read?"
- "Run the weekly report"
Each question maps to one or two MCP tool calls. The agent picks the right tool, runs the query, and returns the answer in chat β no dashboard tab, no copy-paste.
## Why an MCP server for analytics
A dashboard answers the questions you anticipated when you built it. An MCP server lets the agent answer the ones you didn't. Same data, different access pattern: the agent can compose tools, follow up on its own answers, and run the diagnostic queries you'd manually click through.
Tools are typed (`outputSchema` per MCP protocol 2025-06-18), so client-side validation works before any call lands. Per-tool descriptions name the use case and a representative example, so agents pick the right tool on the first try.
## Setup
For one-click install in Cursor / VS Code / Claude Code / Claude Desktop, use the [Clamp dashboard](https://clamp.sh/dashboard). For stdio (CI, headless agents, custom clients, self-hosted instances), pick your client below.
### VS Code / Copilot
Add to `.vscode/mcp.json` in your project:
```json
{
"servers": {
"clamp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@clamp-sh/mcp"],
"env": {
"CLAMP_API_KEY": "sk_proj..."
}
}
}
}
```
### Claude Code
```bash
claude mcp add clamp -- npx -y @clamp-sh/mcp
export CLAMP_API_KEY="sk_proj..."
```
### Claude Desktop and Cursor
Both clients use the same shape β drop this into `claude_desktop_config.json` or `.cursor/mcp.json`:
```json
{
"mcpServers": {
"clamp": {
"command": "npx",
"args": ["-y", "@clamp-sh/mcp"],
"env": {
"CLAMP_API_KEY": "sk_proj..."
}
}
}
}
```
### API key
Get your API key from the Clamp dashboard under **Settings β API Keys**. Keys are scoped to a project and start with `sk_proj`.
## Tools
### Free (all plans)
| Tool | What it does |
|---|---|
| `traffic.overview` | Pageviews, visitors, sessions, bounce rate, avg duration. Period comparison included. |
| `traffic.timeseries` | Any core metric over time with automatic granularity (visitors, sessions, bounce rate, avg duration, or event counts). |
| `traffic.compare` | Compare any metric across two arbitrary periods. Returns absolute and percent delta. |
| `traffic.live` | Visitors active in the last N minutes plus their top pages, referrers, and countries. |
| `breakdown` | Group visitors and pageviews by any dimension: `pathname`, `referrer_host`, `country`, `city`, `region`, `device_type`, `browser`, `browser_version`, `os`, `os_version`, `entry_page`, `exit_page`, `channel`, or any UTM field. |
| `pages.engagement` | Per-page metrics with `view`: `summary`, `engagement` (adds engagement seconds and bounce rate), `sections` (per-section view counts for one pathname; needs the section-views SDK extension). |
| `events.list` | Custom event counts with property filtering and grouping. |
| `events.observed_schema` | The actual fired-event signature with per-property type observations. Diff against a local `event-schema.yaml` to surface schema drift. |
| `events.property_values` | Top distinct string-typed values a property has taken on a specific event. Use to discover the value space before defining a cohort filter or running `events.list` with `property=`. |
| `revenue.sum` | Sum Money-typed event properties. Split by currency, group by any dimension, switch to first-touch via `attribution_model`. Drill-in filters: `plan`, `product`, `first_touch_dim` + `first_touch_value`. |
| `revenue.summary` | The whole Revenue tab in one call: revenue per currency, customers, orders, AOV, LTV, MRR (running balance) + ARR, per-plan rollup (where `mrr_contribution` is the cumulative MRR balance per plan, not the in-period delta), per-product rollup, prior-period comparison, and mode flags (`is_subscription`, `has_one_time`) so callers can render conditionally. |
| `revenue.retention` | Cohort revenue retention on MRR-on-MRR math. Per window: `mature_size`, `retained`, `retention_rate`, `mrr_at_d`, `baseline_mrr_mature`, `nrr` (net revenue retention; >1.0 = net expansion). Anchored at the customer's first `subscription_started`. |
| `revenue.timeseries` | Plot one revenue metric per bucket. `metric` accepts `revenue` (Money sum), `mrr` (running balance, picks up from prior period), `customers` (distinct paying), or `transactions` (event count). |
| `sessions.paths` | Aggregate session paths: top entry β exit pairs with pages per session and duration. |
| `users.journey` | Chronological session-and-event reconstruction for one anonymous ID. |
| `cohorts.create` / `cohorts.list` / `cohorts.retention` / `cohorts.compare` / `cohorts.delete` | Define cohorts by event + period + filter; query retention curves; compare 2β10 cohorts side-by-side on the same retention windows; delete a saved cohort by name (definitions are immutable, so delete + re-create is the rename/restructure flow). |
| `errors.list` / `errors.groups` / `errors.timeline` / `errors.context` | Recent errors, fingerprint-grouped errors with affected-user counts, error rate over time, and breadcrumbs leading to a single error. |
| `projects.list` | List all projects this credential can access. |
| `docs.search` | Keyword-search the Clamp docs index. |
### Pro
| Tool | What it does |
|---|---|
| `funnels.create` | Define and immediately evaluate a multi-step conversion funnel. Steps accept property predicates: `cta_click[location=hero][plan=pro]`. Names must be unique within a project β a duplicate-name call returns 409. |
| `funnels.list` | List funnels or fetch one with cohort filters (country, channel, device, UTM). |
| `funnels.delete` | Delete a saved funnel by name (definitions are immutable, so delete + re-create is the rename/restructure flow). |
| `alerts.create` | Set up metric alerts (e.g. "visitors drops_by 30% over 7d"). |
| `alerts.list` / `alerts.delete` | List and remove alerts. |
### Common parameters
**period** β `"today"`, `"yesterday"`, `"7d"`, `"30d"`, `"90d"`, or a custom range as `"YYYY-MM-DD:YYYY-MM-DD"`. Defaults to `"30d"`.
**limit** β Max rows returned, 1-50. Defaults to 10.
**Filters** β Most tools accept `pathname`, `referrer_host`, `channel`, `country`, `region`, `city`, `device_type`, `browser`, `browser_version`, `os`, `os_version`, and the full UTM set (`utm_source`, `utm_medium`, `utm_campaign`, `utm_content`, `utm_term`). Funnels accept the same cohort filters.
**dimension** (`breakdown` only) β see the `breakdown` row above for valid values.
**view** (`pages.engagement` only) β `"summary"` (default), `"engagement"`, or `"sections"`. `"sections"` requires `pathname`.
**Revenue filters** (`revenue.sum`, `revenue.timeseries`) β `plan` and `product` filter to one subscription tier or one-time product. `first_touch_dim` (one of `channel`, `referrer_host`, `utm_campaign`, `utm_source`, `utm_medium`, `country`) + `first_touch_value` filter to customers whose earliest session matched that dim. `attribution_model="first_touch"` on `revenue.sum` groups by first-touch dim instead of conversion-time dim.
**Revenue metric** (`revenue.timeseries` only) β `"revenue"` (default), `"mrr"` (running balance), `"customers"`, or `"transactions"`.
## Prompts
Pre-built analytics workflows the AI can follow. Each prompt tells the model which tools to call, in what order, and how to present the results.
| Prompt | What it produces |
|---|---|
| `weekly_report` | Traffic summary, top pages, referrers, countries, one actionable insight. |
| `traffic_diagnosis` | Root-cause analysis for traffic changes. Drills into channels, countries, devices, pages. |
| `conversion_audit` | Funnel drop-off analysis with cohort segmentation. |
| `channel_breakdown` | Traffic quality comparison across sources (volume vs engagement). |
| `page_performance` | Deep-dive on a single page: trends, referrers, devices, engagement vs site average. |
## Examples
### Traffic overview
> "How's my traffic this week?"
Calls `traffic.overview` with `period="7d"`. Returns pageviews, visitors, sessions, bounce rate, avg duration, and comparison to the previous week.
> "Show me visitor trends for the last 90 days"
Calls `traffic.timeseries` with `period="90d"` and `metric="visitors"`. Returns daily counts over the full window.
### Sources and geography
> "Where are my visitors coming from?"
Calls `breakdown` with `dimension="referrer_host"`. Returns referrer hostnames with channel classification (organic_search, direct, referral, paid, email, organic_social) joined into each row.
> "Which cities in Germany have the most visitors?"
Calls `breakdown` with `dimension="city"` and `country="DE"`. Returns city names with country and visitor/pageview counts.
> "Is Safari 17 bouncing harder than Chrome 120?"
Two calls to `traffic.overview` with `browser="Safari"`, `browser_version="17"` and `browser="Chrome"`, `browser_version="120"`. Returns side-by-side bounce rates.
> "Which pages did Hacker News visitors read?"
Calls `pages.engagement` with `referrer_host="news.ycombinator.com"`. Returns only pages viewed by traffic from HN.
### Custom events
> "How many signups happened this month?"
Calls `events.list` with `name="signup"` and `period="30d"`.
> "Break down signups by plan"
Calls `events.list` with `name="signup"` and `group_by="plan"`. Returns counts per property value.
### Funnels (Pro)
> "Create a funnel from pricing page to signup to purchase"
Calls `funnels.create` with steps `["pageview:/pricing", "signup", "purchase"]`. Returns step-by-step conversion rates immediately.
> "Compare funnel conversion for US vs Germany"
Two calls to `funnels.list`, one with `country="US"` and one with `country="DE"`.
### Alerts (Pro)
> "Alert me if visitors drop by 30% week over week"
Calls `alerts.create` with `metric="visitors"`, `condition="drops_by"`, `threshold=30`, `period="7d"`.
### Workflows
> "Why did traffic drop last week?"
Follows the `traffic_diagnosis` prompt. Systematically checks channels, countries, devices, and pages to isolate the root cause.
> "Audit our conversion funnel"
Follows the `conversion_audit` prompt. Analyzes funnel drop-offs, segments by device/country/channel, and identifies the biggest optimization opportunity.
## Environment variables
| Variable | Required | Description |
|---|---|---|
| `CLAMP_API_KEY` | Yes | Project API key (`sk_proj...`). |
| `CLAMP_API_URL` | No | Override the API base. Defaults to `https://api.clamp.sh`. Set this when pointing at a self-hosted Clamp instance. |
## FAQ
**Do I need an API key for the Remote MCP install?** No. Installing from the Clamp dashboard handles auth automatically. The API key is only needed for stdio (this npm package), self-hosted instances, CI, and custom clients.
**How is this different from the Clamp dashboard?** Same data, different access pattern. Use the dashboard when you want a chart; use the MCP when you want the agent to do the digging β diagnose a drop, compose a funnel, run a weekly report β and return the answer in chat.
**Can I self-host?** Yes. Point `CLAMP_API_URL` at your self-hosted Clamp instance, or run the stdio server inline in your own infrastructure. The MCP tools work against any Clamp API endpoint that speaks the same protocol.
**Does this work for read-only audits, or can the agent make changes?** Both. The Free-tier tools are all read-only except for `cohorts.create` / `cohorts.delete`; the Pro tools (`funnels.create`, `funnels.delete`, `alerts.create`, `alerts.delete`) also write. Scope the connection from the dashboard if you only want read access.
**Will the agent know about my custom events?** Yes. `events.observed_schema` returns the actual fired-event signatures with per-property type observations, so the agent can answer "what's in my event schema" without you pasting anything.
## License
MIT
