MCP App Store
by KeenEthics
Riley icon

Riley MCP Server for Claude and ChatGPT

By Riley

Overview

Claude is incredibly powerful—but without business context, it’s like Google Maps without GPS. It can answer questions, but it doesn’t know your customers, revenue, sales process, or what matters most to your business. Riley transforms fragmented revenue data into the live context Claude needs to become an expert on your company. Connect Salesforce, HubSpot, Gong, Slack, customer emails, product usage, support tickets, documents, call transcripts, and CRM data into a continuously updated intelligence layer. Riley cleans, connects, and understands this information so Claude reasons over your business. Ask Claude: • Which accounts are most likely to expand next quarter? • Which deals are off playbook? • Why is pipeline slowing in healthcare? • What customer themes emerged this week? • Which competitors are appearing more frequently? • What actions should Customer Success take to improve NRR? Unlike dashboards that describe what happened, Riley continuously discovers patterns, identifies risks, surfaces opportunities, and keeps Claude updated as your business changes. Claude is the reasoning engine. Riley is the Revenue Context Layer. Together, they function like GPS for your revenue organization, helping Sales, Customer Success, Marketing, Product, and executives make faster, better decisions based on a complete, continuously evolving understanding of the business

How to connect

  1. Claude: Open the Riley listing in the Claude connectors directory and choose Connect. Claude listing Riley exposes 8 tools here.
  2. ChatGPT: Open the Riley listing in the ChatGPT app directory and choose Connect. ChatGPT listing Riley exposes 11 tools here.

Tools

riley_analyze

ChatGPT
Run a natural language analysis against the customer's Riley data. A demo workspace with no data of its own is analyzed against the shared Kora sample datasets and any health scores created on top of them. Use this for customer and business analytics questions about customer health, churn risk, accounts at risk, renewals, revenue, product usage, support risk, expansion, or account trends. Also use it for requests phrased as 'According to Riley'. For example: 'What are the top 5 accounts at risk of churning?' Returns immediately with a processing handle (analysis_result_id, poll_after_seconds) — the analysis itself runs on Riley's backend. Call riley_get_analysis_status with the returned analysis_result_id, waiting poll_after_seconds between calls, until status is 'completed', 'failed', or 'timed_out'. Polling is capped at 5 minutes; past that, riley_get_analysis_status returns status='timed_out' and the LLM should stop polling. Completed results include citation stubs only; call riley_get_citation to load a citation's full table.

riley_confirm_ingest_upload

ChatGPT
Pairs with riley_request_ingest_upload (backfill / bulk path). Call after uploading the file with the pre-signed URL. Riley scans + validates + promotes the file. Returns the job's current state; poll riley_get_ingest_status(job_id) until status is one of: promoted, rejected_format, rejected_malware, rejected_schema, failed, expired. Not needed for riley_inline_ingest_upload — that path starts scanning automatically.

riley_create_dataset

ChatGPT
Create a new dataset for this customer. Declare columns with SQL types (string, integer, number, boolean, date, timestamp, or SQL aliases like VARCHAR / BIGINT / DOUBLE / TIMESTAMP). Optionally pass primary_key for row-level updates (later ingests with the same key replace prior values) or omit it for append-only ingest. After creation, push records with riley_request_ingest_upload using the returned dataset_definition_id.

riley_get_analysis_status

ChatGPT
Fetch the status (and final result, if complete) of a riley_analyze call. Use this to poll an analysis_result_id returned by riley_analyze with status=='processing'. Stop polling when status becomes 'completed', 'failed', or 'timed_out' (5-minute polling cap). Completed payloads include citation stubs (title, columns, a small preview) rather than full tables. Call riley_get_citation to load a citation's full table.

riley_get_citation

ChatGPT
Return the full row table for one citation from a completed riley_analyze result, in a single call. Do not pass offset, limit, or other paging arguments. Completed analyses include citation stubs (title, columns, a 5-row preview) rather than the full table — call this when the preview is not enough. Check rows_analyzed on the stub first; large citations can be tens or hundreds of thousands of rows. Prefer the preview unless the user needs the full set.

riley_get_ingest_status

ChatGPT
Returns the current state of an ingest job (works for both riley_inline_ingest_upload and riley_request_ingest_upload). Poll until status is one of: promoted, rejected_format, rejected_malware, rejected_schema, failed, expired. Restricted to customer admins.

riley_get_persona

ChatGPT
Get full detail for one persona — motivations, behaviors, frictions, signals they care about, evidence, and representative quotes. Same data as the Persona Detail page. Pass persona_id (from riley_list_personas) or name (case-insensitive; matches exactly, or a unique substring).

riley_inline_ingest_upload

ChatGPT
Daily-sync / chat-driven path. Upload a small file (≤10 MB raw, ~14 MB base64) by sending its bytes inline as base64 in this JSON-RPC call. Use this when the agent runtime can't issue an arbitrary HTTP PUT (e.g., Claude Desktop). For larger files or programmatic bulk ingests use riley_request_ingest_upload (pre-signed PUT) instead. Files are partitioned by date received (today UTC). Returns the job in scanning state; poll riley_get_ingest_status until terminal. Pass idempotency_key to make retries safe.

riley_list_datasets

ChatGPT
List datasets Riley can analyze for this customer. A demo workspace with no data of its own returns the shared Kora sample datasets (accounts, usage, support tickets, call notes, and opportunities); those rows are read-only. Also returns completed health-score definitions (scores array) with the virtual column names to use in riley_analyze — e.g. riley__churn_risk / riley__churn_risk_category. Prefer those score columns over CRM health fields when assessing churn or account health. Returns each dataset's dataset_definition_id, name, description, and whether it supports upsert-by-key. Pass include_columns=true to see physical column names and types (score columns are listed under scores, not under dataset columns). Physical datasets that are members of a combined (merged) dataset are omitted; use the combined dataset for analysis. Do not ingest into a dataset whose read_only flag is true.

riley_list_personas

ChatGPT
List behavioral personas Riley has discovered for this customer, from the latest completed persona snapshot (same data as the Persona Hub page). Returns each persona's id, name, tagline, source (defined/emerging), and entity_count. Read-only — call before riley_get_persona to pick the right persona_id or name.

riley_request_ingest_upload

ChatGPT
Backfill / bulk path. Returns a pre-signed PUT URL the agent uploads its file to (CSV, JSONL/NDJSON, or Parquet), plus a job_id. Then call riley_confirm_ingest_upload(job_id) to start processing. Use this when the agent runtime can issue an HTTP PUT (CLI, SDK, Cursor agent, Code Interpreter) — for chat clients without HTTP capability (Claude Desktop), use riley_inline_ingest_upload instead. Maximum file size: 500 MB per upload (oversized files are rejected post-PUT; pass expected_bytes to fail fast at this step instead). Files are partitioned by date received (today UTC). Prefer one large upload over many small ones — per-row uploads are extremely inefficient. Pass idempotency_key to make retries safe.

riley_analyze

Claude

riley_confirm_ingest_upload

Claude

riley_create_dataset

Claude

riley_get_analysis_status

Claude

riley_get_ingest_status

Claude

riley_inline_ingest_upload

Claude

riley_list_datasets

Claude

riley_request_ingest_upload

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/riley"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/riley)