Amplitude MCP Tools
Amplitude MCP gives your AI agent tools across the Amplitude product suite. This reference groups tools by capability. To connect a client, refer to the client setup guides. For prompts and multi-step workflows, refer to Amplitude MCP use cases.
Tool availability
Your client, data region, connected products, authentication method, and Amplitude permissions determine the tools and actions available. Region-specific and product-specific requirements appear with the relevant tools below.
ChatGPT uses a curated tool catalog by default and exposes search in place of the two search tools listed here. For a manually configured connection, progressive discovery lets the client discover additional available tools. Discovery doesn't grant extra access.
Write operations require the Use MCP (write) permission and any permissions required by the underlying product. Service accounts created through Settings provide read access.
Context and discovery
| Tool | Capability |
|---|---|
get_amplitude_context | Identify your organization, user, and accessible projects. Read project settings and organization or project AI Context. |
manage_amp_context | Replace the AI Context text for an organization or project. |
search_amp_entities | Find saved Amplitude content, including charts, dashboards, notebooks, experiments, cohorts, and metrics. |
get_from_url | Read the object referenced by a supported Amplitude URL. |
Start a session with get_amplitude_context so your agent applies your team's definitions and business rules. It returns AI Context text. Uploaded reference files provide titles and matching snippets rather than complete file contents. Configure these in AI Controls.
For public documentation search and retrieval, connect the separate Docs MCP server.
Analytics and charts
| Tool | Capability |
|---|---|
query_amplitude_data | Run ad hoc analytics and query saved charts, including segmentation, funnels, and retention. |
get_amplitude_charts | Read saved chart definitions, configuration, and data. |
render_amplitude_chart | Display an interactive Amplitude chart. |
use_amplitude_chart_monitors | Create and manage chart monitors, alerts, and alert history. |
Dashboards, notebooks, and collaboration
| Tool | Capability |
|---|---|
use_amp_dashboards | Read, create, and edit dashboards, including titles, descriptions, charts, sections, and layout. |
use_amp_notebooks | Read, create, and edit notebooks containing charts, rich text, and multi-column layouts. |
use_amp_comments | Read and add comments on Amplitude content. |
share_amp_entities | Manage sharing and access roles for supported Amplitude content. |
manage_amp_entities | Create or update shared spaces and move charts, dashboards, notebooks, and cohorts into them. |
Dashboard edits save referenced temporary chart edits as permanent charts. Dashboard and notebook edits check for concurrent changes. Shared-space changes require confirmation before they apply.
Users, cohorts, and metrics
| Tool | Capability |
|---|---|
get_amp_user_data | Read product-user profiles, properties, and activity timelines. |
use_amplitude_cohorts | Read and create cohorts, inspect membership, export CSVs, and manage destination syncs and their history. |
use_amplitude_metrics | Read, create, update, and delete metric definitions, and manage metric goals and goal alerts. |
Cohort syncs show a preview and require confirmation before sending membership to a configured destination.
Experiments and feature flags
| Tool | Capability |
|---|---|
use_amp_experiments | Read, analyze, create, and update feature experiments, including their metrics and variant results. |
use_amp_flags | Read, create, and update flags, variants, testers, and deployments for flags and feature experiments. |
Taxonomy and tracking plans
| Tool | Capability |
|---|---|
search_amp_data_taxonomy | Find events, properties, and property values using a natural-language description. |
get_amp_taxonomy | Read tracking-plan definitions, deleted items, branch differences, activity history, governance settings, environments, and branches. |
manage_amp_events | Create, update, delete, and restore eligible events, custom events, and labeled events. |
manage_amp_properties | Create event, user, and group properties; update, delete, and restore eligible event and user properties. |
get_transformations | Read event merges, property merges, and property-value mappings configured in the project. |
get_group_types | List group categories such as Company, Team, or Account. |
manage_amp_branches | Create, update, review, refresh, merge, and delete tracking-plan branches. |
These tools change tracking-plan definitions. Send production event data through an Amplitude SDK or the HTTP V2 API.
Event and property changes require Use MCP (write) and Update Tracking Plan. Custom-event changes require Manage Data Enrichments instead of Update Tracking Plan. For delete and restore examples, refer to Clean up your taxonomy.
Session Replay
| Tool | Capability |
|---|---|
get_amp_session_replay_info | Find and list recordings, then inspect their recorded interactions and analytics-event timelines. |
AI Visibility
| Tool | Capability |
|---|---|
use_amp_ai_visibility | Read tracked-brand visibility, topics, prompts, competitors, citations, and sentiment; configure measurement events and properties. |
Customer feedback
| Tool | Capability |
|---|---|
use_amplitude_ai_feedback | Explore feedback themes, search comments, inspect supporting feedback, and review connected sources and saved trends. |
AI agents and Agent Analytics
| Tool | Capability |
|---|---|
get_agent_results | Read results from agents that analyzed dashboards or replays. Available in the US data center only. |
get_amplitude_agent_analytics_info | Analyze AI-agent performance, quality, cost, sessions, spans, conversations, schemas, and available filter values. |
To change Agent Analytics objects, use run_amp with the supported agent-analytics commands.
| Object | Capability through run_amp |
|---|---|
| Datasets | Create or update datasets and add or remove sessions. |
| Evaluators | Create or update an evaluator. |
| Evaluation runs | Trigger runs and retrieve or list their results. |
Removing dataset sessions requires allowDestructive: true. Project access and credential scopes apply to every call.
Guides and Surveys
Guides and surveys don't have a dedicated tool. Use run_amp with guides or surveys commands to read configuration and create or edit drafts. These capabilities are available in the US and EU regions.
Commands through run_amp | Capability |
|---|---|
guides list, surveys list | List items and filter by platform, creator, creation date, or archived status. |
guides get, surveys get | Read steps, content variants, targeting, triggers, and publication details. |
guides create, surveys create | Create an unpublished draft and return its Amplitude admin URL. |
guides update, surveys update | Edit draft metadata, content, page targeting, and supported trigger and lifecycle settings. |
guides archive, surveys archive | Archive an unpublished item, with destructive-operation opt-in. |
Finish in Amplitude: set audience targeting and rollout, publish, schedule, or test the guide or survey in the UI. Changes made through MCP remain drafts until published there.
Guides and Surveys reads aren't supported through enterprise-managed auth. For scopes, command syntax, response fields, draft editing limits, and example calls, refer to Manage guides and surveys with Amplitude MCP.
Data warehouse and connections
Read configured warehouse and cloud-storage connections and inspect import and export jobs. The published catalog uses a consolidated tool in the US and individual source/job tools in the EU.
| Tool | Capability |
|---|---|
use_amplitude_data_connections | In the US, list sources, read source details and job history, check connection health, and browse connected Snowflake warehouses. |
get_data_ingestion_sources | In the EU, list project ingestion sources, including SDKs, warehouse imports, and integrations, with status and recent volume. |
get_data_source_details | In the EU, read configuration and metadata for an ingestion source. |
get_data_warehouse_jobs | In the EU, inspect import or export job history for one connection, including manual and backfill exports. |
get_data_warehouse_destinations | List export destinations, their status, and recent export timestamps. This remains separate from the consolidated tool. |
For connection inputs, job parameters, status values, response fields, and export destination timestamps, refer to Data warehouse MCP tools.
Opportunities and Wave
Wave is a paid offering for Growth and Enterprise organizations.
| Tool | Capability |
|---|---|
query_wave_opportunities | Find and read Wave opportunities and their details. |
query_wave_product_areas | Find and read Wave product areas. |
manage_wave_opportunities | Create, update, and relate opportunities. |
manage_wave_product_areas | Create and update product areas. |
manage_wave_verification_artifacts | Attach and manage opportunity verification artifacts. |
Additional capabilities with the Developer CLI
| Tool | Capability |
|---|---|
run_amp | Run supported Developer CLI commands through MCP for additional capabilities, including guides, surveys, heatmaps, zoning, and Agent Analytics objects. |
Ask for command help before choosing a command or flags. Put only the subcommand in command and options without leading dashes in flags. The session fixes the host, region, and credential. Delete and archive commands require allowDestructive: true.
For example, { "command": ["help", "guides"] } lists the supported guide commands. Refer to the Developer CLI quickstart and Developer API for further reference.
For reusable workflow playbooks, refer to the Amplitude MCP Marketplace and its open-source repository.
Tool discovery
In progressive mode, your client starts with a small discovery catalog and uses these helpers to inspect the tools available to the connection.
| Tool | Capability |
|---|---|
list_tool_categories | List available product categories. |
get_category_tools | Find tools in a category. |
describe_tool | Read a tool's current schema and required parameters before calling it. |
For connection URLs and the discovery sequence, refer to Progressive tool discovery.
Was this helpful?
For AI agents: a documentation index is available at /docs/llms.txt. Append .md to any page URL for markdown, or send Accept: text/markdown.