- Access is the annotation Claude uses for permissions. Read-only tools run without asking. Tools that change data are marked destructive, so Claude asks before each call.
- Scope is the OAuth scope your token needs. See Scopes.
Platform overview
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_agent_health | Read-only | mcp:read | Get comprehensive health status of a specific agent: recent executions, error rate, and metrics. |
get_platform_overview | Read-only | mcp:read | Get a high-level overview of the platform: total agents, recent executions, errors, and general health status. |
get_recent_activity | Read-only | mcp:read | Get the most recent activity across the platform: latest executions and deployments. |
Documentation
| Tool | Access | Scope | What it does |
|---|---|---|---|
fetch_doc_page | Read-only | mcp:read | Fetch the full content of a Runflow documentation page by URL. |
search_docs | Read-only | mcp:read | Search the Runflow SDK documentation for guides and code examples. |
Agents
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_agent | Read-only | mcp:read | Get detailed information about a specific agent including stats and configuration |
get_agent_business_metrics | Read-only | mcp:read | Get business KPIs for an agent based on its category (Sales: conversion rate, leads, meetings; Support: resolution time, CSAT, tickets; Voice: call duration, sentiment; etc.). |
get_agent_code | Read-only | mcp:read | Download the source code and files of an agent. |
get_agent_executions | Read-only | mcp:read | Get recent execution history for an agent. |
get_agent_tree | Read-only | mcp:read | Get the file tree structure of an agent repository. |
get_category_metrics | Read-only | mcp:read | Get aggregated business metrics for all agents in a category (e.g. |
list_agents | Read-only | mcp:read | List all agents in the tenant with optional pagination |
chat_with_agent | Changes data | mcp:write | Send a message to a deployed Runflow agent and get its response. |
create_agent | Changes data | mcp:write | Create a new AI agent with the specified configuration |
delete_agent | Changes data | mcp:admin | Delete an agent permanently |
deploy_agent | Changes data | mcp:write | Deploy an agent with file changes to STAGING. |
promote_agent | Changes data | mcp:write | Promote agent from staging to production. |
update_agent | Changes data | mcp:write | Update an existing agent configuration |
update_agent_file | Changes data | mcp:write | Update a single file in an agent repository (STAGING environment). |
Executions
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_execution | Read-only | mcp:read | Get full details of a specific execution: all traces, spans, tool calls, LLM inputs/outputs, token usage, cost, and timing. |
get_thread_executions | Read-only | mcp:read | Get all executions within a specific conversation thread. |
list_executions | Read-only | mcp:read | List recent agent executions (conversation threads). |
Observability, reviews, events and dashboards
| Tool | Access | Scope | What it does |
|---|---|---|---|
aggregate_events | Read-only | mcp:read | Aggregate values from the event store, returns a single number (KPI) or a time/category series. |
check_execution_has_review | Read-only | mcp:read | Check whether a given execution already has a review attached. |
export_execution_reviews_for_training | Read-only | mcp:write | Export reviews as a dataset for LLM fine-tuning (OpenAI conversational format). |
get_dashboard_card | Read-only | mcp:read | Get a single dashboard card by ID. |
get_dashboard_cards | Read-only | mcp:read | [Deprecated alias of list_dashboard_cards] List all dashboard cards for an agent. |
get_event_names | Read-only | mcp:read | List event names the agent has logged. |
get_event_properties | Read-only | mcp:read | List available property keys for a specific event. |
get_execution_details | Read-only | mcp:read | Get detailed trace information for a specific execution including all spans and steps |
get_execution_metrics | Read-only | mcp:read | Infrastructure-level metrics: total executions, success/failure counts, token usage, cost, duration. |
get_execution_review | Read-only | mcp:read | Fetch a single execution review by its ID, including input/output snapshots and resolution metadata. |
get_execution_reviews_stats | Read-only | mcp:read | Aggregate counts for execution reviews: total, by status (pending/in_progress/resolved), by rating (bad/needs_improvement), by priority (critical/high), plus average resolution time in hours. |
list_dashboard_cards | Read-only | mcp:read | List all dashboard metric cards for an agent. |
list_dashboard_tabs | Read-only | mcp:read | List dashboard tabs for an agent, sorted by sortOrder. |
list_execution_reviews | Read-only | mcp:read | List execution reviews (agent QA feedback) with filters. |
list_execution_threads | Read-only | mcp:read | List execution threads (conversations/sessions) with status and timing. |
query_events | Read-only | mcp:read | Table-style query against the event store. |
render_dashboard_card | Read-only | mcp:read | Resolve a dashboard card to its current value by reusing its polymorphic config. |
create_dashboard_card | Changes data | mcp:write | Create a dashboard metric card. |
create_dashboard_tab | Changes data | mcp:write | Create a new dashboard tab. |
create_execution_review | Changes data | mcp:write | Create a review for a specific execution. |
delete_dashboard_card | Changes data | mcp:admin | Delete a dashboard card permanently. |
delete_dashboard_tab | Changes data | mcp:admin | Delete a dashboard tab. |
delete_execution_review | Changes data | mcp:admin | DESTRUCTIVE: permanently delete a review row. |
move_dashboard_card | Changes data | mcp:write | Move a dashboard card to a different tab. |
reorder_dashboard_cards | Changes data | mcp:write | Bulk-update card positions in a single transaction. |
reorder_dashboard_tabs | Changes data | mcp:write | Bulk-update tab sort order in a single transaction. |
update_dashboard_card | Changes data | mcp:write | Update a dashboard card, any subset of fields can be updated. |
update_dashboard_tab | Changes data | mcp:write | Rename a tab and/or change its sortOrder. |
update_execution_review | Changes data | mcp:write | Update a review, change its status, attach resolution notes, record the action taken (e.g. |
Sessions
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_session_history | Read-only | mcp:read | Get the conversation history for a session |
list_sessions | Read-only | mcp:read | List conversation sessions with filters. |
clear_session | Changes data | mcp:write | Clear all messages from a session |
Prompts
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_prompt | Read-only | mcp:read | Get a prompt by name or ID, optionally resolved for a specific environment |
list_prompts | Read-only | mcp:read | List all prompts in the tenant |
render_prompt | Read-only | mcp:read | Render a prompt template with variables. |
create_prompt | Changes data | mcp:write | Create a new prompt template |
delete_prompt | Changes data | mcp:admin | Delete a prompt by name or ID. |
update_prompt | Changes data | mcp:write | Update a prompt by creating a new STAGING draft version. |
Connectors
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_connector | Read-only | mcp:read | Get a connector by slug with its full details: instance params (baseUrl, etc), credential, and all available resources (HTTP actions). |
list_connector_instances | Read-only | mcp:read | List configured connector instances in the tenant |
list_connector_templates | Read-only | mcp:read | List available connector templates (prebuilt integrations catalog) |
add_connector_resource | Changes data | mcp:write | Add a new resource (HTTP action) to an existing CUSTOM connector. |
add_resource_to_template | Changes data | mcp:write | Add a resource (HTTP action) to a CUSTOM connector template that has NO instances yet (setup phase, before create_connector_instance). |
create_connector_instance | Changes data | mcp:write | Create a new connector instance from a template. |
create_connector_resource | Changes data | mcp:write | Create a custom resource (endpoint) for a REST API or Database connector instance. |
create_connector_template | Changes data | mcp:write | Create a new CUSTOM connector template in the tenant catalog. |
create_custom_connector | Changes data | mcp:write | Create a reusable custom connector in the tenant catalog in ONE call. |
delete_connector_instance | Changes data | mcp:admin | Delete a STAGING connector instance. |
delete_connector_resource | Changes data | mcp:admin | Remove a resource from a CUSTOM connector. |
execute_connector | Changes data | mcp:write | Execute a connector resource (make an API call through a connector). |
update_connector_instance | Changes data | mcp:write | Update a STAGING connector instance, change name or credential. |
update_connector_resource | Changes data | mcp:write | Update an existing resource on a CUSTOM connector (change method, path, description, or any schema). |
Credentials
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_credential | Read-only | mcp:read | Get a credential by name or ID |
list_credentials | Read-only | mcp:read | List all credentials configured in the tenant |
create_credential | Changes data | mcp:write | Create a new credential (API key, bearer token, basic auth, OAuth2, etc.) |
delete_credential | Changes data | mcp:admin | Delete a credential by ID |
test_credential | Changes data | mcp:write | Test if a credential is valid and can connect |
update_credential | Changes data | mcp:write | Update a credential, change name, value, or description. |
Knowledge bases
| Tool | Access | Scope | What it does |
|---|---|---|---|
list_documents | Read-only | mcp:read | List documents in a vector store |
list_vector_stores | Read-only | mcp:read | List all vector stores (knowledge bases) in the tenant |
search_knowledge | Read-only | mcp:read | Perform semantic search across a vector store (knowledge base) |
add_document | Changes data | mcp:write | Add a text document to a vector store for future semantic search. |
create_vector_store | Changes data | mcp:write | Create a new vector store (knowledge base) for document storage and semantic search |
delete_document | Changes data | mcp:admin | Delete a document from a vector store |
delete_vector_store | Changes data | mcp:admin | Delete a vector store and all its documents. |
update_vector_store | Changes data | mcp:write | Update a vector store, change name or description. |
Data sources
| Tool | Access | Scope | What it does |
|---|---|---|---|
list_datasources | Read-only | mcp:read | List all datasources configured in the tenant |
create_datasource | Changes data | mcp:write | Create a new datasource connection |
delete_datasource | Changes data | mcp:admin | Delete a datasource. |
test_datasource | Changes data | mcp:write | Test a datasource connection |
update_datasource | Changes data | mcp:write | Update a datasource, change name, type, or connection string. |
Triggers
| Tool | Access | Scope | What it does |
|---|---|---|---|
list_triggers | Read-only | mcp:read | List all triggers (cron jobs, webhooks, schedulers) in the tenant |
create_trigger | Changes data | mcp:write | Create a new trigger for an agent. |
delete_trigger | Changes data | mcp:admin | Delete a trigger. |
toggle_trigger | Changes data | mcp:write | Activate or deactivate a trigger. |
update_trigger | Changes data | mcp:write | Update a trigger, change name, config, or status. |
LLM providers
| Tool | Access | Scope | What it does |
|---|---|---|---|
get_available_provider_types | Read-only | mcp:read | List all available LLM provider types with metadata (OpenAI, Anthropic, Bedrock, Azure OpenAI, Groq, Gemini, Custom). |
get_compatible_credentials | Read-only | mcp:read | List credentials that are compatible with a specific LLM provider type. |
get_llm_provider | Read-only | mcp:read | Get detailed information about a specific LLM provider including its configuration |
list_llm_providers | Read-only | mcp:read | List all configured LLM providers (OpenAI, Anthropic, Bedrock, Groq, Gemini, etc.) with their status and model count |
list_provider_models | Read-only | mcp:read | List all models available for a specific LLM provider. |
create_llm_provider | Changes data | mcp:write | Configure a new LLM provider. |
delete_llm_provider | Changes data | mcp:admin | Delete an LLM provider configuration and all its models |
set_default_llm_provider | Changes data | mcp:write | Set an LLM provider as the default for its type. |
test_llm_provider | Changes data | mcp:write | Test if an LLM provider connection is working. |
Releases
| Tool | Access | Scope | What it does |
|---|---|---|---|
list_pending_releases | Read-only | mcp:read | List all pending changes between staging and production (prompts with drafts, agents with staging commits, connectors with changes) |
publish_to_production | Changes data | mcp:write | Publish staging changes to production. |