Each action is granted on its own. An agent allowed to read is not thereby allowed to write, and the scope beside each row is what the acting user must have connected for it to run at all.
nango_delete_connections_by_connectionidWRITE
Permanently delete an end-user connection and the third-party credential Nango holds for it. Syncs for that connection stop and its records are removed; the end user must re-authorise to restore it.
environment:connections:delete
nango_delete_integrations_by_uniquekeyWRITE
Permanently delete an integration and every connection that belongs to it. The end users behind those connections will have to re-authorise.
environment:integrations:delete
nango_delete_integrations_by_uniquekey_functions_by_nameWRITE
Delete one deployed function from an integration. A function that was deployed by the Nango CLI cannot be deleted here -- the API answers `function_managed_by_deploy`.
environment:functions:delete
nango_delete_sessions_by_sessionidWRITE
End an agent session immediately, revoking its session token and MCP URL. Anything the agent was mid-way through stops having access.
environment:agent_sessions:write
nango_get_action_by_actionidREAD
Fetch the result of an action that was triggered with run_async. Poll this with the id trigger_action returned until the action has finished.
environment:actions:execute
nango_get_connectionsREAD
List the end-user connections in this Nango environment, optionally filtered by connection id, integration, end user or free text. Returns each connection's id, integration, provider, end user, tags and any active errors -- never the stored third-party credential.
environment:connections:list
nango_get_functions_invocations_by_invocationidREAD
Fetch the result of a function invocation by its id, for invocations that were queued rather than awaited.
environment:functions:invocations
nango_get_integrationsREAD
List the integrations configured in this Nango environment -- the customer's own connected apps, each with its unique_key, provider, display name, logo and webhook-forwarding setting.
environment:integrations:list
nango_get_integrations_by_uniquekeyREAD
Fetch one integration by its unique key: provider, display name, webhook-forwarding setting, and which credential and connection-config fields are already set at the integration level (names only, never values).
environment:integrations:read
nango_get_integrations_by_uniquekey_functionsREAD
List the sync, action and on-event functions deployed for one integration, paged and filterable by type or name.
environment:functions:list
nango_get_integrations_by_uniquekey_functions_by_nameREAD
Fetch one deployed function's definition: its type, trigger, models, frequency and input/output schemas.
environment:functions:read
nango_get_integrations_by_uniquekey_functions_by_name_codeREAD
Return the TypeScript source of one deployed function. Useful for reviewing or debugging what a sync actually fetches.
environment:functions:read
nango_get_providersREAD
List every API provider Nango can integrate with (1000+), with each provider's auth mode, categories, proxy base URL and default scopes. This is Nango's catalogue, not the customer's own integrations -- use nango_get_integrations for those. Requires no key scope.
no provider scope
nango_get_providers_by_providerREAD
Fetch one provider's full configuration from Nango's catalogue: auth mode, authorization and token URLs, proxy base URL, available and default scopes, and any connection-configuration fields its integrations require. Requires no key scope.
no provider scope
nango_get_providers_by_provider_templatesREAD
List the pre-built sync and action function templates Nango publishes for one provider, with each template's name, description, required scopes and input/output schema. Requires no key scope.
no provider scope
nango_get_proxy_by_anypathREAD
Issue a GET against the third-party API behind one of the customer's connections, with Nango attaching (and refreshing) the stored credential server-side. The credential is never returned to the caller. Reads only -- the proxy's write verbs are not exposed. What a GET does is the destination API's business, though: a few APIs mutate on GET.
environment:proxy
nango_get_recordsREAD
Read the records one sync has written for a connection, as a cursor-paged list. This is how synced third-party data is consumed: filter by change type or modification time to fetch only what moved since the last read.
environment:records:read
nango_get_scripts_configREAD
Return every sync and action function deployed in this environment, with their models, schedules and input/output schemas -- the whole integration surface in one read. `format=openai` returns the actions as OpenAI function definitions.
environment:integrations:list_functions
nango_get_sync_statusREAD
Report the status of one or more syncs for a connection: schedule state, last run result, next run times and record counts. The first place to look when data is stale.
environment:syncs:read
nango_patch_connections_by_connectionidWRITE
Update a connection's end-user record, tags, or webhook URL override. Does not touch the stored credential.
environment:connections:update
nango_patch_connections_metadataWRITE
MERGE keys into the metadata stored on one or more connections, leaving keys you do not mention untouched. Use set_connection_metadata to replace the whole object.
environment:connections:update
nango_patch_integrations_by_uniquekeyWRITE
Rename an integration, change its display name, or change whether it forwards webhooks. Credentials are not accepted here -- change them in the Nango dashboard.
environment:integrations:update
nango_patch_records_pruneWRITE
Permanently delete synced records for a connection up to a cursor. Used to reclaim storage on a model that has grown; the records are gone and only a full re-sync brings them back.
environment:records:write
nango_post_action_triggerWRITE
Run one of the integration's action functions against a connection and return its result -- the write path for a Nango integration (create a contact, send a message). Set run_async to queue it and poll get_action_result instead of waiting.
environment:actions:execute
nango_post_connect_sessionsWRITE
Mint a short-lived Connect session for one end user and return a `connect_link` they can open to authorise an integration. This is the supported way to create a connection: the end user's own credential is entered in Nango's hosted UI and never passes through this tool.
environment:connect_sessions:write
nango_post_connect_sessions_reconnectWRITE
Mint a Connect session that re-authorises an EXISTING connection whose credential has expired or been revoked, and return the link for the end user to open. Keeps the connection's id, syncs and records.
environment:connect_sessions:write
nango_post_connections_metadataWRITE
REPLACE the metadata stored on one or more connections. Metadata is the arbitrary JSON a sync or action function reads at runtime (the field list to fetch, a cursor, a customer-specific setting). This overwrites whatever was there -- use update_connection_metadata to merge.
environment:connections:update
nango_post_functions_invocationsWRITE
Invoke an already-deployed function by name against a connection and return its result. The newer, general form of trigger_action -- it takes a function name and input, never source code.
environment:functions:invocations
nango_post_integrationsWRITE
Create a new integration in this Nango environment for one of the providers in Nango's catalogue. Credentials for the integration are NOT accepted here -- set the OAuth client id/secret in the Nango dashboard.
environment:integrations:create
nango_post_integrations_quickstartWRITE
Create an integration pre-wired with Nango's own shared OAuth credentials for the provider, so it can be connected without registering a developer app first. Intended for evaluation; use create_integration plus the dashboard for production credentials.
environment:integrations:create
nango_post_sessionsWRITE
Create a Nango agent session and return its MCP URL and session token, scoped to a chosen set of the customer's connections and tools. This is how an AI agent is given time-boxed, toolset-limited access to the third-party APIs Nango holds credentials for.
environment:agent_sessions:write
nango_post_sync_pauseWRITE
Pause the schedule of one or more syncs. Existing records stay; no new runs happen until start_sync resumes them.
environment:syncs:execute
nango_post_sync_startWRITE
Resume the SCHEDULE of one or more paused syncs, so they run on their frequency again. Use trigger_sync for a single run now.
environment:syncs:execute
nango_post_sync_triggerWRITE
Run one or more syncs now, outside their schedule. By default the run is incremental; `opts.reset` re-fetches everything from scratch.
environment:syncs:execute
nango_put_sync_update_connection_frequencyWRITE
Override how often one sync runs for one connection, or clear the override by passing a null frequency to fall back to the sync's own schedule.
environment:syncs:update