Python bindings for the Quicknode SDK.

This is one of four language bindings published from the same Rust core. See the project README for the polyglot overview, development setup, and release process.

Pre-1.0: While on 0.x, releases may contain breaking changes. Check the release notes before upgrading.

• Installation
• Quick Start
• Configuration
  • Option A — Pass config directly
  • Option B — Load from environment (from_env())
  • Custom headers and User-Agent
• Platform Support
• API Reference
  • Language conventions
  • Admin Client
    • Endpoints
    • Endpoint Tags
    • Teams
    • Usage
    • Logs
    • Endpoint Security
    • Security Options
    • Tokens
    • Referrers
    • IPs
    • Domain Masks
    • JWTs
    • Request Filters
    • Multichain
    • IP Custom Headers
    • Method Rate Limits
    • Endpoint Rate Limits
    • Endpoint URLs
    • Metrics
    • Chains
    • Account
    • Billing
    • Bulk Operations
    • Account Tags
    • Additional security operations
    • Mutation responses
    • Usage and metrics summaries
    • Activity logs
    • Transactions
    • Admin compatibility notes
  • Streams Client
    • Datasets, Regions, and Destinations
    • Streams methods
  • Webhooks Client
    • Templates and destination
    • Webhooks methods
  • KV Store Client
    • Sets
    • Lists
  • SQL Client
  • RPC & Tooling Access
• Crypto-micropayment lane (rpc.call)
  • Wallet generation
  • x402 credit drawdown (authenticate once, then draw one credit per call)
    • Testnet faucet
  • MPP payment channel (deposit once, then vouchers)
• Error Handling
• License

uv add quicknode-sdk

Construct the SDK once, then reach into the five sub-clients (admin, streams, webhooks, kvstore, sql). Subsequent API Reference snippets assume you have a qn handle from one of these blocks.

# Python
import asyncio
from quicknode_sdk import QuicknodeSdk

async def main():
    qn = QuicknodeSdk.from_env()
    resp = await qn.admin.get_endpoints()
    print(f"{len(resp.data)} endpoints")

asyncio.run(main())

There are two ways to configure the SDK.

# Python
from quicknode_sdk import QuicknodeSdk, SdkFullConfig, HttpConfig
qn = QuicknodeSdk(SdkFullConfig(api_key="your-key", http=HttpConfig(timeout_secs=30)))

# api_key is optional: the crypto-micropayment lane pays per request instead, so
# api_key=None builds a usable SDK. Every other client still needs one, and
# from_env() always requires QN_SDK__API_KEY.

# Python
qn = QuicknodeSdk.from_env()

Environment variables (prefix QN_SDK__, separator __):

Every outbound HTTP request includes an auto-generated User-Agent of the form:

quicknode-sdk-<language>/<sdk-version> (<os>-<arch>; <language>-<runtime-version>)

You can attach arbitrary headers via HttpConfig.headers. These headers OVERRIDE any SDK-managed header with the same name, including User-Agent, x-api-key, Accept, and Content-Type. Use this to inject correlation IDs, proxy auth, or to replace the default User-Agent. Header names are matched case-insensitively.

from quicknode_sdk import QuicknodeSdk, SdkFullConfig, HttpConfig

qn = QuicknodeSdk(
    SdkFullConfig(
        api_key="your-key",
        http=HttpConfig(headers={
            "X-Correlation-Id": "abc-123",
            "User-Agent": "my-app/1.0",  # overrides SDK default
        }),
    )
)

Precompiled wheels are published for:

Linux glibc wheels are built against glibc 2.17 so they load on any distro from 2014 onward — RHEL 7+, Ubuntu 14.04+, Debian 8+, Amazon Linux 2+, SLES 12+, Fedora 19+. If pip install quicknode-sdk resolves to a source distribution on your platform, you're on something we don't have a prebuilt wheel for — see the matrix above.

Not supported: RHEL/CentOS 6 (glibc 2.12), Debian 7 (glibc 2.13), Ubuntu 12.04 (glibc 2.15), SLES 11 (glibc 2.11), Intel macOS, Windows.

Snippets assume qn was already constructed via the Quick Start. Optional parameters are skipped unless showing one is needed to illustrate usage.

• Methods are async — call with await. Parameters are kwargs; responses are native pyclass objects with attribute access.

Accessed as qn.admin. Manages endpoints, tags, teams, billing, usage, metrics, security, and rate limits. Backed by https://api.quicknode.com/v0/.

Returns a paginated list of endpoints on the account with optional search, filters (networks, statuses, labels, tags, dedicated, flat-rate), sorting, and pagination. The API caps limit at 100. Nonempty tag_ids and tag_labels filters are mutually exclusive.

Parameters (all optional): limit (i32), offset (i32), search (string), sort_by (string), sort_direction ("asc" | "desc"), networks (string[]), statuses (string[]), labels (string[]), dedicated (bool), is_flat_rate (bool), tag_ids (i32[]), tag_labels (string[]).

Returns: GetEndpointsResponse — { data: Endpoint[], pagination?: Pagination }.

# Python
resp = await qn.admin.get_endpoints(limit=20, sort_by="created_at", sort_direction="desc")

Creates a new endpoint for the given blockchain and network.

Parameters: chain and network are strings required by the API. The SDK retains its existing optional arguments for compatibility; supply both for a valid request.

Returns: CreateEndpointResponse with data: SingleEndpoint.

# Python
resp = await qn.admin.create_endpoint(chain="ethereum", network="mainnet")

Fetches a single endpoint by id, including its full security configuration and rate limits.

Parameters: id (string, required).

Returns: ShowEndpointResponse with data: SingleEndpoint.

# Python
resp = await qn.admin.show_endpoint("ep-123")

Updates editable fields on an endpoint. Currently supports label.

Parameters: id (string, required); label (string, required by the API). The SDK retains the optional label argument; omitting it produces an API error.

Returns: nothing.

# Python
await qn.admin.update_endpoint("ep-123", label="my label")

Archives an endpoint. The HTTP verb is DELETE but the effect is archival, not permanent deletion.

Parameters: id (string, required).

Returns: nothing.

# Python
await qn.admin.archive_endpoint("ep-123")

Pauses or unpauses an endpoint.

Parameters: id (string, required); body: status (string, required — "active" or "paused").

Returns: UpdateEndpointStatusResponse.

# Python
await qn.admin.update_endpoint_status("ep-123", status="paused")

Per-endpoint tag add/remove. For account-wide tag management see Account Tags.

Tags an endpoint with the given label. Creates the tag on the account if it does not exist.

Parameters: id (string, required); label (string, required by the API). Labels must be nonblank and contain 3 through 25 characters after trimming. The SDK retains its optional argument and does not trim values.

Returns: nothing.

# Python
await qn.admin.create_tag("ep-123", label="prod")

Removes a tag from a specific endpoint.

Parameters: id (endpoint id, string, required), tag_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_tag("ep-123", "42")

Lists all teams on the account.

Parameters: none.

Returns: ListTeamsResponse with data: TeamSummary[].

# Python
resp = await qn.admin.list_teams()

Creates a new team. The name must contain at least four characters.

Parameters: name (string, required).

Returns: CreateTeamResponse with data: CreateTeamData.

# Python
resp = await qn.admin.create_team(name="Payments")

Fetches team detail including pending invites.

Parameters: id (i64, required).

Returns: GetTeamResponse with data: TeamDetail.

# Python
resp = await qn.admin.get_team(42)

Deletes a team.

Parameters: id (i64, required).

Returns: DeleteTeamResponse.

# Python
await qn.admin.delete_team(42)

Lists endpoints accessible to a team.

Parameters: id (i64, required).

Returns: ListTeamEndpointsResponse with data: TeamEndpoint[].

# Python
resp = await qn.admin.list_team_endpoints(42)

Replaces the set of endpoints associated with a team. Pass an empty array to remove all.

Parameters: id (i64, required); body: endpoint_ids (string[], required).

Returns: UpdateTeamEndpointsResponse.

# Python
await qn.admin.update_team_endpoints(42, endpoint_ids=["ep-123", "ep-456"])

Invites a user to a team. Existing users only need email; new users require full_name and role.

Parameters: id (i64, required); body: email (string, required), full_name (string, optional), role (string, optional). Standard roles and account custom-role keys are accepted.

Returns: InviteTeamMemberResponse.

# Python
await qn.admin.invite_team_member(42, email="alice@example.com", role="viewer")

Removes a user from a team.

Parameters: id (team id, i64, required), user_id (i64, required).

Returns: RemoveTeamMemberResponse.

# Python
await qn.admin.remove_team_member(42, 7)

Re-sends a pending team invitation.

Parameters: id (team id, i64, required), user_id (i64, required).

Returns: ResendTeamInviteResponse.

# Python
await qn.admin.resend_team_invite(42, 7)

Endpoint, chain, and tag usage rows include integer rpc_credits, data_credits, and bytes. Shared method rows, including methods nested under endpoints, include integer bytes. Existing nullable method metadata and null tag IDs for untagged usage remain nullable.

All usage methods accept optional start_time and end_time Unix timestamps. Omit both to use the current billing-period default.

Aggregate account usage for a time window.

Returns: GetUsageResponse with data: UsageData (credits_used, credits_remaining, limit, overages, start_time, end_time).

# Python
resp = await qn.admin.get_usage()

Per-endpoint usage breakdown.

Returns: GetUsageByEndpointResponse with data.endpoints: EndpointUsage[].

# Python
resp = await qn.admin.get_usage_by_endpoint()

Per-RPC-method usage breakdown.

Returns: GetUsageByMethodResponse with data.methods: MethodUsage[].

# Python
resp = await qn.admin.get_usage_by_method()

Per-chain usage breakdown.

Returns: GetUsageByChainResponse with data.chains: ChainUsage[].

# Python
resp = await qn.admin.get_usage_by_chain()

Per-tag usage breakdown.

Returns: GetUsageByTagResponse with data.tags: TagUsage[].

# Python
resp = await qn.admin.get_usage_by_tag()

Fetches a page of request logs for an endpoint. Use ISO 8601 timestamps for explicit bounds. The limit is 1 through 100 and defaults to 20. Set include_details=true for full request/response payloads (truncated at 2 KB each).

Parameters: id (endpoint id, required); query: from (string timestamp, required), to (string timestamp, required), include_details (bool, optional), limit (i32, optional), next_at (string cursor, optional).

Returns: GetEndpointLogsResponse — { data: EndpointLog[], next_at?: string }.

# Python
resp = await qn.admin.get_endpoint_logs(
    "ep-123",
    from_time="2026-04-01T00:00:00Z",
    to_time="2026-04-02T00:00:00Z",
    limit=100,
)

Returns the full request/response payloads for a single log entry.

Parameters: id (endpoint id, required), request_id (log request uuid, required).

Returns: GetLogDetailsResponse with data: LogDetails.

# Python
resp = await qn.admin.get_log_details("ep-123", "req-abc")

Returns the full security configuration for an endpoint: tokens, JWTs, referrers, domain masks, IPs, blocked IPs (blocked_ips), request filters, and their per-feature toggles. Snapshot options include hsts, cors, and response_logging (wire key responseLogging). Structured filter matchers and params are preserved.

Parameters: id (string, required).

Returns: GetEndpointSecurityResponse with data: EndpointSecurity.

# Python
resp = await qn.admin.get_endpoint_security("ep-123")

Returns the list of security features and their enabled state for an endpoint.

Parameters: id (string, required).

Returns: GetSecurityOptionsResponse with data: SecurityOption[].

# Python
resp = await qn.admin.get_security_options("ep-123")

Enables or disables individual security features. Each field accepts "enabled" or "disabled".

Parameters: id (string, required); options: SecurityOptionsUpdate (tokens, referrers, jwts, ips, domain_masks, hsts, cors, request_filters, ip_custom_header, response_logging). The response_logging option is sent as responseLogging and accepts "enabled" or "disabled".

Returns: UpdateSecurityOptionsResponse with updated SecurityOption[].

# Python
await qn.admin.update_security_options(
    "ep-123",
    tokens="enabled",
    jwts="disabled",
    response_logging="enabled",
)

Generates a new auth token on an endpoint.

Parameters: id (endpoint id, required).

Returns: nothing.

# Python
await qn.admin.create_token("ep-123")

Revokes a token on an endpoint.

Parameters: id (endpoint id, required), token_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_token("ep-123", "tok-1")

Whitelists a referrer URL or domain on an endpoint.

Parameters: id (endpoint id, required); body: referrer (string, required).

Returns: nothing.

# Python
await qn.admin.create_referrer("ep-123", referrer="example.com")

Removes a referrer from the whitelist.

Parameters: id (endpoint id, required), referrer_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_referrer("ep-123", "ref-1")

Whitelists an IP address on an endpoint.

Parameters: id (endpoint id, required); body: ip (string, required).

Returns: nothing.

# Python
await qn.admin.create_ip("ep-123", ip="198.51.100.7")

Removes an IP from the whitelist.

Parameters: id (endpoint id, required), ip_id (string, required).

Returns: DeleteBoolResponse.

# Python
await qn.admin.delete_ip("ep-123", "ip-1")

Adds a custom domain mask to an endpoint.

Parameters: id (endpoint id, required); body: domain_mask (string, required by the API). The SDK retains its optional argument, but supplies no default value.

Returns: nothing.

# Python
await qn.admin.create_domain_mask("ep-123", domain_mask="rpc.example.com")

Removes a domain mask.

Parameters: id (endpoint id, required), domain_mask_id (string, required).

Returns: DeleteBoolResponse.

# Python
await qn.admin.delete_domain_mask("ep-123", "dm-1")

Configures JWT validation on an endpoint.

Parameters: id (endpoint id, required); body: public_key (string, required), kid (string, required), name (string, required).

Returns: nothing.

# Python
await qn.admin.create_jwt(
    "ep-123",
    public_key="-----BEGIN PUBLIC KEY-----\n...",
    kid="key-1",
    name="primary",
)

Removes a JWT configuration.

Parameters: id (endpoint id, required), jwt_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_jwt("ep-123", "jwt-1")

Whitelist specific RPC methods on an endpoint. Requests for methods not on the list are blocked when the feature is enabled.

Parameters: id (endpoint id, required); body: method (string[], required). Ruby's Hash key is methods (plural).

Returns: CreateRequestFilterResponse with data.id.

# Python
resp = await qn.admin.create_request_filter(
    "ep-123",
    method=["eth_blockNumber", "eth_getBalance"],
)

Parameters: id (endpoint id, required), request_filter_id (string, required); body: method (string[], optional). Ruby's Hash keys are request_filter_id and methods (plural).

Returns: nothing. Supply a method list for an update; an empty request produces an API error.

# Python
await qn.admin.update_request_filter("ep-123", "f-1", method=["eth_call"])

Parameters: id (endpoint id, required), request_filter_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_request_filter("ep-123", "f-1")

get_request_filters returns list records with id, pattern, and method_whitelist. This differs from endpoint/security snapshots, where each filter has id, a method matcher array, and params. Snapshot params holds every top-level pattern key except method, not only a nested params matcher.

Use create_request_filter_pattern and update_request_filter_pattern for a custom JSON object. The existing whitelist methods keep their current inputs, including Ruby's methods: keyword. Do not combine a whitelist and a custom pattern in one request or use an update to switch filter kinds. One whitelist and multiple custom filters can coexist. Updates and deletions continue to accept successful empty HTTP 204 responses.

Nested objects, arrays, nulls, and keys such as anything-but remain intact. Pass a native object rather than a JSON string. Creation returns data.id and, when supplied by the API, data.pattern.

pattern = {"method": [{"prefix": "eth_"}], "jsonrpc": ["2.0"]}
created = await qn.admin.create_request_filter_pattern("123", pattern=pattern)
if created.data is not None:
    await qn.admin.update_request_filter_pattern("123", created.data.id, pattern=pattern)
filters = await qn.admin.get_request_filters("123")

Enables multichain on an endpoint.

Parameters: id (endpoint id, required).

Returns: nothing.

# Python
await qn.admin.enable_multichain("ep-123")

Disables multichain on an endpoint.

Parameters: id (endpoint id, required).

Returns: nothing.

# Python
await qn.admin.disable_multichain("ep-123")

Sets the custom header used to identify the client IP (e.g. when traffic is proxied).

Parameters: id (endpoint id, required); body: header_name (string, required).

Returns: CreateOrUpdateIpCustomHeaderResponse with data.header_name.

# Python
await qn.admin.create_or_update_ip_custom_header("ep-123", header_name="X-Forwarded-For")

Removes the custom IP header configuration.

Parameters: id (endpoint id, required).

Returns: DeleteBoolResponse.

# Python
await qn.admin.delete_ip_custom_header("ep-123")

Lists method-level rate limiters configured on an endpoint.

Parameters: id (endpoint id, required).

Returns: GetMethodRateLimitsResponse with data.rate_limiters: MethodRateLimiter[].

# Python
resp = await qn.admin.get_method_rate_limits("ep-123")

Creates a new method-level rate limiter. rate must be positive; zero produces an API error. Optional limit_by_ip applies the limit per client IP and is returned on MethodRateLimiter.

Parameters: id (endpoint id, required); body: interval (string, e.g. "second"), methods (string[]), rate (i32), limit_by_ip (bool, optional).

Returns: CreateMethodRateLimitResponse with data: MethodRateLimiter.

# Python
resp = await qn.admin.create_method_rate_limit(
    "ep-123",
    interval="second",
    methods=["eth_call"],
    rate=10,
    limit_by_ip=True,
)

Updates an existing rate limiter. Only provided fields change. Supply at least one supported field; an empty PATCH produces an API error. limit_by_ip=false is an explicit update, not an omitted value.

Parameters: id (endpoint id, required), method_rate_limit_id (string, required); body: methods (string[], optional), status ("enabled" | "disabled", optional), rate (i32, optional), limit_by_ip (bool, optional).

Returns: UpdateMethodRateLimitResponse.

# Python
await qn.admin.update_method_rate_limit("ep-123", "rl-1", rate=50, limit_by_ip=False)

Deletes a rate limiter.

Parameters: id (endpoint id, required), method_rate_limit_id (string, required).

Returns: nothing.

# Python
await qn.admin.delete_method_rate_limit("ep-123", "rl-1")

Partial update of the endpoint-level RPS / RPM / RPD caps. Only supplied fields change. Values are capped by the account's plan tier. Sends PATCH. Set limit_by_ip (limitByIp in Node) to true to apply the caps per client IP or false to apply them to the whole endpoint. You can update this flag without sending caps. Supply at least one cap or the flag; an empty update returns an API error. Both update methods accept these options.

Parameters: id (endpoint id, required); rps, rpm, rpd: optional integers; limit_by_ip: optional boolean.

Returns: nothing.

# Python
await qn.admin.update_rate_limits("ep-123", rps=100, rpm=5000, limit_by_ip=True)
await qn.admin.update_rate_limits("ep-123", limit_by_ip=False)

Returns the rate-limit rows currently enforced on the endpoint, each identifying its bucket ("rps" / "rpm" / "rpd"), rate_limit, and source ("plan_default" or "user_override"). User-set overrides expose an id (camelCased id in Node) you can pass to delete_rate_limit_override.

Parameters: id (endpoint id, required).

Returns: GetRateLimitsResponse with data.rate_limits: RateLimitEntry[].

# Python
resp = await qn.admin.get_rate_limits("123")
for row in resp.data.rate_limits:
    print(row.bucket, row.rate_limit, row.source, row.id)

Deletes a user-set rate-limit override by UUID. Plan defaults are not deletable — passing a UUID that does not match a user-set override on the endpoint returns 404.

Parameters: id (endpoint id, required); override_id / overrideId (UUID returned by get_rate_limits, required).

Returns: nothing.

# Python
await qn.admin.delete_rate_limit_override("123", "ovr-uuid")

Returns the HTTP and WebSocket URLs for the endpoint without fetching the full endpoint record. For multichain endpoints, multichain_urls / multichainUrls is a per-network map of additional URLs; for single-chain endpoints it is None / null.

Parameters: id (endpoint id, required).

Returns: GetEndpointUrlsResponse with data.http_url, data.wss_url, and data.multichain_urls.

# Python
resp = await qn.admin.get_endpoint_urls("123")
print(resp.data.http_url)
if resp.data.multichain_urls:
    for network, urls in resp.data.multichain_urls.items():
        print(network, urls.http_url)

Returns metric series for an endpoint over a time period. Optional integer percentile ranges from 1 through 100 and is required for response_time_over_time and method_response_time_max. For example, use 95. Metric and period strings remain open.

Parameters: id (endpoint id, required); query: period ("hour" | "day" | "week" | "month"), metric (e.g. "method_calls_over_time", "response_status_breakdown"), percentile (integer, optional).

Returns: GetEndpointMetricsResponse with data: list[EndpointMetric]. Each EndpointMetric has a tag: list[str] and a data: list[list[int]] of [timestamp, value] pairs. Single-axis series (e.g. response_time_over_time with a percentile) come back as a one-element tag like ["p95"]; multi-axis series come back as ["network", "arbitrum-mainnet"].

# Python
resp = await qn.admin.get_endpoint_metrics(
    "ep-123",
    period="day",
    metric="response_time_over_time",
    percentile=95,
)

Returns account-level metric series. The existing optional percentile argument remains for compatibility; the API currently ignores it. Endpoint latency percentile support belongs to get_endpoint_metrics.

Parameters: period (required), metric (required), percentile (string, optional).

Returns: GetAccountMetricsResponse with data: list[EndpointMetric]. See get_endpoint_metrics above for the tag: list[str] shape.

# Python
resp = await qn.admin.get_account_metrics(period="day", metric="credits_over_time")

Lists the blockchains supported by Quicknode along with their networks.

Parameters: none.

Returns: ListChainsResponse with data: Chain[].

# Python
resp = await qn.admin.list_chains()

Returns details about the account, including its id, name, creation timestamp, billing version, and current subscription.

Parameters: none.

Returns: AccountInfoResponse with data: AccountInfo (including a nested subscription: AccountSubscription).

# Python
resp = await qn.admin.account_info()

Returns the per-method API credit costs for a chain, identified by its slug (the same slugs returned by list_chains, e.g. ethereum). An unknown chain slug raises ApiError (status 404).

Parameters: chain (string, required) — the chain slug.

Returns: GetApiCreditsResponse with data: list[ApiCredit], where each ApiCredit has method and credits.

# Python
resp = await qn.admin.get_api_credits("ethereum")

Lists invoices on the account.

Parameters: none.

Returns: ListInvoicesResponse with data.invoices: Invoice[].

# Python
resp = await qn.admin.list_invoices()

Lists payments on the account.

Parameters: none.

Returns: ListPaymentsResponse with data.payments: Payment[].

# Python
resp = await qn.admin.list_payments()

Each result can include an error string. Check individual results: HTTP 200 can contain both successes and failures. Successful rows can omit error.

Activates or pauses many endpoints at once.

Parameters: ids (string[], required), status ("active" | "paused", required).

Returns: BulkUpdateEndpointStatusResponse with per-endpoint results.

# Python
resp = await qn.admin.bulk_update_endpoint_status(ids=["ep-1", "ep-2"], status="paused")

Applies a tag (created if missing) to many endpoints at once.

Parameters: ids (string[], required), label (string, required).

Returns: BulkAddTagResponse.

# Python
resp = await qn.admin.bulk_add_tag(ids=["ep-1", "ep-2"], label="prod")

Removes a tag from many endpoints at once.

Parameters: ids (string[], required), tag_id (i32, required).

Returns: BulkRemoveTagResponse.

# Python
resp = await qn.admin.bulk_remove_tag(ids=["ep-1", "ep-2"], tag_id=42)

Lists every tag on the account along with usage counts.

Parameters: none.

Returns: ListTagsResponse with data.tags: AccountTag[].

# Python
resp = await qn.admin.list_tags()

Renames an account-level tag. Labels must be nonblank and contain 3 through 25 characters after trimming. The SDK does not trim values.

Parameters: tag_id (i32, required); body: label (string, required).

Returns: RenameTagResponse with updated AccountTag.

# Python
resp = await qn.admin.rename_tag(42, label="staging")

Deletes a tag from the account. The tag must first be removed from any endpoints using it.

Parameters: id (i32, required).

Returns: DeleteAccountTagResponse.

# Python
await qn.admin.delete_account_tag(42)

All methods below take the numeric endpoint id as a string. Reads do not enable a security feature.

Certificate records contain fingerprint, subject, issuer, and expires_at. Deleting the last certificate returns disabled state with an empty list; DELETE still has a response body. Invalid, expired, duplicate, or unknown certificates can return HTTP 422.

Blacklist and range-limit updates replace the complete configuration. An empty array clears the blacklist; an empty object clears range limits. Range limits must be positive integers. Read supported_methods for the endpoint's supported methods instead of assuming one global list. Blocked-IP calls do not toggle IP security; visibility can depend on the endpoint's IP security setting.

certificates = await qn.admin.get_client_certs("123")
ranges = await qn.admin.get_range_limits("123")
blocked = await qn.admin.get_blocked_ips("123")
blacklist = await qn.admin.update_method_blacklist("123", methods=["eth_sendRawTransaction"])

Existing mutation methods keep their original return values. Use these response-returning variants when you need the created record or effective limits. Each takes the same arguments as its existing counterpart.

Creation and snapshot fields differ: a created referrer uses url, while endpoint snapshots use referrer; a created domain mask uses domain_mask, while snapshots use domain. Tag association id differs from tag_id; these IDs are numeric. Rust Debug output redacts token and JWT credential values. Store returned credentials securely and do not log them.

created = await qn.admin.create_referrer_with_response("123", referrer="https://example.com")
print(created.data.id)

These methods return a resolved time window, typed totals, optional grouped rows, and an optional continuation cursor.

Error breakdown rows contain kind and count, plus the selected grouping dimensions. Other breakdown rows carry their summary metrics and selected dimensions.

For the first page, supply last as a positive integer duration ending now (s, m, h, or d), or both from_time and to_time as ISO 8601 timestamps with timezones. Bare dates are not summary timestamps. Use group_by for grouping. Credits summaries have no grouping parameter. The limit is 1 through 1000 and defaults to 100.

Pass native arrays for groups and array filters. The SDK encodes repeated bracketed query keys such as group_by[]=method&group_by[]=endpoint. Error codes and HTTP status-code filters remain strings. endpoint_names contains endpoint subdomains; RPC, traffic, and account-error result endpoint values are integer endpoint IDs. Account-error endpoints filters accept integer arrays, such as [123, 456]. tag_ids and addon_ids contain numeric IDs represented as strings. Endpoint error-summary id uses the numeric endpoint ID represented as a string.

For error summaries, kind is rpc or http. Method filtering/grouping requires rpc; RPC kind rejects status_codes, and HTTP kind rejects codes. Feature access and retention can produce API errors, including HTTP 403.

Continue with cursor alone. A cursor belongs to the original route, account, and query; any repeated parameters, including limit, must match. Keep the returned window values: the server can align summary windows to billing boundaries. Totals cover the full window on every page, not only the current rows. The SDK does not round windows or paginate automatically.

Summary metrics use floating-point numbers, including count-like fields, and can contain fractions such as 7.5. They do not promise arbitrary decimal precision. Legacy usage fields remain integers. RPC breakdown requests and traffic breakdown successful_requests can be null even though totals are non-null. A missing breakdown is not an empty grouped result.

Endpoint grouping values are int or None. For a grouping dimension, an omitted attribute raises AttributeError; a selected null dimension returns None. Use hasattr(row, "method") to check presence. An omitted breakdown returns None. Summary window attributes are from_time and to_time.

page = await qn.admin.get_usage_rpc_summary(
    last="1h", group_by=["method", "endpoint"], limit=20,
)
print(page.data.totals.credits)
if page.data.next_cursor is not None:
    next_page = await qn.admin.get_usage_rpc_summary(cursor=page.data.next_cursor)

Lists account changes, newest first. list_activity_log_filters returns the actor IDs and target types accepted by list_activity_logs; get_activity_log returns one entry by UUID.

List filters are q, actor, endpoint_id, target_type, from_time, to_time, limit, and offset. endpoint_id is a numeric endpoint ID represented as a string. target_type is sent as type. The limit is 0 through 100 and defaults to 20. Offset pagination returns total, limit, and offset; pin to_time when exporting because new entries shift offsets. Bare dates are accepted: from_time starts at midnight UTC, and to_time includes the entire day.

Responses keep actor and target metadata typed. Actor IDs, names, and emails can be null, including for deleted users. Feature access can return HTTP 403; a missing detail entry returns HTTP 404.

filters = await qn.admin.list_activity_log_filters()
page = await qn.admin.list_activity_logs(limit=20)
if page.data:
    detail = await qn.admin.get_activity_log(page.data[0].id)

list_transactions returns retained transactions and an optional continuation cursor. A first page needs last (for example, 1h), or both from_time and to_time as ISO 8601 timestamps with timezones. A cursor alone continues the same resolved window.

Optional filters are network, method, state, transaction_hash, and endpoints. States are submitted, succeeded, and failed. Endpoint filters accept integer endpoint IDs, such as [123, 456], and use repeated endpoints[] query keys. The limit is 1 through 1000 and defaults to 100.

get_transaction takes a network, a transaction hash, and optional include_payloads. It returns an outer data containing exists and a second data value. An absent or expired transaction can return exists: false and data: null. Payloads default to excluded; set include_payloads to true to request retained request and response text. The SDK preserves that text without parsing it. Feature access can return HTTP 403.

Transaction endpoint_ids arrays contain integer endpoint IDs or null elements; the detail array itself can be absent or null. Each attempt has a required integer endpoint_id. Block heights remain decimal strings, including values such as 9007199254740993. Other detail metadata can be absent or null. Attempts retain the order returned by the API.

page = await qn.admin.list_transactions(last="1h", endpoints=[123, 456], limit=20)
if page.data.transactions:
    transaction = page.data.transactions[0]
    result = await qn.admin.get_transaction(
        transaction.network, transaction.transaction_hash, include_payloads=False,
    )
    if result.data.exists and result.data.data is not None:
        print(result.data.data.height)  # str or None
if page.data.next_cursor is not None:
    next_page = await qn.admin.list_transactions(cursor=page.data.next_cursor)

Summary endpoint dimensions and transaction endpoint IDs use int values. Pass integer lists to account-error and transaction endpoints filters. Transaction endpoint-ID lists can contain None; attempt endpoint_id is required. Update callers that used UUID strings.

Existing methods retain their return values and optional argument requiredness. Required API inputs are documented above; omitted or invalid values still produce API errors. Use the response-returning methods when you need created records.

Rust callers must update affected full struct literals for the added fields, even when those fields are optional. This includes security snapshots/options, filter responses, usage rows, bulk results, method-rate-limit requests/responses, and endpoint metric requests. Request structs with Default can use ..Default::default() for omitted fields; builders avoid full-literal updates where available.

EndpointRequestFilter.method changed from Vec<String> to Vec<serde_json::Value> so it can preserve structured matchers. Convert strings with .into() or serde_json::json!, and use as_str() when reading a string matcher. Match object values as JSON rather than assuming every matcher is a string. Snapshot params and custom pattern are object maps containing JSON values.

Python endpoint/security response wrappers keep their public class names, getters, and setters. Filter values are native lists and dictionaries; assign through the setter after editing a returned value. Node uses unknown[] for matcher arrays and Record<string, unknown> for open pattern objects. Ruby continues to return IndifferentHash values. JSON object keys are preserved without recursive case conversion.

New summary, activity, and transaction window fields use from_time and to_time in Rust/Python. Node fields and serialized response keys remain from and to; Python/Ruby client arguments use from_time and to_time. Existing endpoint-log arguments do not change.

Accessed as qn.streams. Creates and manages blockchain data streams that deliver filtered on-chain events to configured destinations. Backed by https://api.quicknode.com/streams/rest/v1/.

Enums used across stream methods:

• StreamRegion: UsaEast, EuropeCentral, AsiaEast (wire values: usa_east, europe_central, asia_east).
• StreamDataset: Block, BlockWithReceipts, Transactions, Logs, Receipts, TraceBlocks, DebugTraces, BlockWithReceiptsDebugTrace, BlockWithReceiptsTraceBlock, BlobSidecars, ProgramsWithLogs, Ledger, Events, Orders, Trades, BookUpdates, Twap, WriterActions.
• StreamStatus: Active, Paused, Terminated, Completed, Blocked.
• FilterLanguage: Javascript, Go, Wasm.
• StreamMetadataLocation: Body, Header, None.

Destinations are expressed via DestinationAttributes. Each variant wraps an attribute struct:

Wrapper naming per language:

• Rust: DestinationAttributes::Webhook(WebhookAttributes { .. }) etc.
• Python: StreamWebhookDestination(WebhookAttributes(...)), StreamS3Destination(S3Attributes(...)), etc.
• Node.js: a discriminated object { destination: "webhook", attributes: { ... } } using string discriminators.
• Ruby: factory methods on QuicknodeSdk::DestinationAttributes, e.g. QuicknodeSdk::DestinationAttributes.webhook(url: ..., ...).

Creates a new stream that delivers filtered data to the configured destination. Start from a specific block for backfills or from the tip for real-time streaming. Supports filters, reorg handling, distance-from-tip, elastic batching, notification emails, and extra destinations.

Parameters: CreateStreamParams — required: name, region, network, dataset, start_range (i64), end_range (i64, -1 = follow tip), destination_attributes, plan, threshold_fetch_buffer. Common optional fields: dataset_batch_size, include_stream_metadata, fix_block_reorgs, keep_distance_from_tip, elastic_batch_enabled, filter_function, filter_language, status, notification_email, extra_destinations.

Returns: Stream.

# Python
from quicknode_sdk import WebhookAttributes, StreamWebhookDestination

stream = await qn.streams.create_stream(
    name="My Stream",
    network="ethereum-mainnet",
    dataset="block",
    region="usa_east",
    start_range=24691804,
    end_range=24691904,
    destination_attributes=StreamWebhookDestination(
        WebhookAttributes(
            url="https://webhook.site/...",
            max_retry=3,
            retry_interval_sec=1,
            post_timeout_sec=10,
            compression="none",
        )
    ),
    plan="growth_plan",
    threshold_fetch_buffer=1000,
    status="active",
)

Paginated list of streams on the account.

Parameters (all optional): offset (i64), limit (i64), order_by (string), order_direction ("asc" | "desc"), stream_type (string).

Returns: ListStreamsResponse with data: Stream[] and page_info.

# Python
resp = await qn.streams.list_streams()

Fetches one stream by id.

Parameters: id (string, required).

Returns: Stream.

# Python
stream = await qn.streams.get_stream("stream-id")

Partially updates a stream. Omitted fields are left unchanged.

Parameters: id (string, required); body: any field from CreateStreamParams (all optional).

Returns: updated Stream.

# Python
stream = await qn.streams.update_stream("stream-id", name="Renamed")

Deletes one stream by id.

Parameters: id (string, required).

Returns: nothing.

# Python
await qn.streams.delete_stream("stream-id")

Deletes every stream on the account. Destructive and takes no arguments.

Parameters: none.

Returns: nothing.

# Python
await qn.streams.delete_all_streams()

Resumes delivery on a stream from its current position.

Parameters: id (string, required).

Returns: nothing.

# Python
await qn.streams.activate_stream("stream-id")

Halts delivery on a stream.

Parameters: id (string, required).

Returns: nothing.

# Python
await qn.streams.pause_stream("stream-id")

Runs a filter function against a block so it can be validated before being attached to a live stream.

Parameters: network (string, required), dataset (StreamDataset, required), block (string, required), filter_function (string, optional), filter_language (FilterLanguage, optional), address_book_config (optional).

Returns: TestFilterResponse with result and logs.

# Python
resp = await qn.streams.test_filter(
    network="ethereum-mainnet",
    dataset="block",
    block="17811625",
)

Counts currently enabled (active) streams, optionally filtered by type.

Parameters: stream_type (string, optional).

Returns: EnabledCountResponse with total.

# Python
resp = await qn.streams.get_enabled_count()

Accessed as qn.webhooks. Creates webhooks from filter templates and manages their lifecycle. Backed by https://api.quicknode.com/webhooks/rest/v1/.

WebhookTemplateId identifies the filter template:

TemplateArgs carries the arguments. Each template supports two input forms — inline values (*Args(*Template(...))) or a reference to a pre-created list by name (*ByListArgs(*ByListTemplate(...))):

WebhookDestinationAttributes: url (required), compression (required — "none" | "gzip"), security_token (optional — auto-generated if omitted).

WebhookStartFrom: Last (resume from last delivered block) or Latest (start from newest).

In Ruby, template_args is passed as a JSON string under the key template_args_json; destination is passed as a JSON string under destination_attributes_json.

Paginated list of webhooks.

Parameters (all optional): limit (i64), offset (i64).

Returns: ListWebhooksResponse with data: Webhook[] and pageInfo: WebhookPageInfo { limit, offset, total }.

# Python
resp = await qn.webhooks.list_webhooks()

Fetches a webhook by id.

Parameters: id (string, required).

Returns: Webhook.

# Python
webhook = await qn.webhooks.get_webhook("wh-1")

Creates a webhook from a predefined filter template.

Parameters: name (required), network (required), destination_attributes (WebhookDestinationAttributes, required), template_args (required — use the TemplateArgs enum variant for the chosen template), notification_email (optional).

Returns: Webhook.

# Python
from quicknode_sdk import EvmWalletFilterArgs, EvmWalletFilterTemplate, WebhookDestinationAttributes

webhook = await qn.webhooks.create_webhook_from_template(
    name="Wallet Webhook",
    network="ethereum-mainnet",
    destination_attributes=WebhookDestinationAttributes(
        url="https://webhook.site/...",
        compression="none",
    ),
    template_args=EvmWalletFilterArgs(
        EvmWalletFilterTemplate(wallets=["0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48"])
    ),
)

Partially updates a webhook's name, notification email, and/or destination. If destination_attributes is supplied without security_token, a new token is generated automatically.

Parameters: id (required); body — all optional: name, notification_email, destination_attributes. In Ruby, destination_attributes is passed as a JSON string under the key destination_attributes_json.

Returns: updated Webhook.

# Python
webhook = await qn.webhooks.update_webhook("wh-1", name="Renamed Webhook")

Updates the template args (and optionally name, email, destination) on an existing template-backed webhook.

Parameters: webhook_id (required), template_args (required); optional: name, notification_email, destination_attributes.

Returns: updated Webhook.

# Python
webhook = await qn.webhooks.update_webhook_template(
    "wh-1",
    template_args=EvmWalletFilterArgs(
        EvmWalletFilterTemplate(wallets=["0xnewwallet"])
    ),
)

Deletes a webhook.

Parameters: id (required).

Returns: nothing.

# Python
await qn.webhooks.delete_webhook("wh-1")

Deletes every webhook on the account. Destructive and takes no arguments.

Parameters: none.

Returns: nothing.

# Python
await qn.webhooks.delete_all_webhooks()

Pauses a webhook so it stops delivering events.

Parameters: id (required).

Returns: nothing.

# Python
await qn.webhooks.pause_webhook("wh-1")

Activates a paused or new webhook so it resumes delivering events. start_from determines where processing resumes.

Parameters: id (required), start_from (WebhookStartFrom, required — Last or Latest).

Returns: nothing.

# Python
await qn.webhooks.activate_webhook("wh-1", start_from="latest")

Counts currently enabled webhooks.

Parameters: none.

Returns: WebhookEnabledCountResponse with total.

# Python
resp = await qn.webhooks.get_enabled_count()

Accessed as qn.kvstore. Provides two primitives — sets (single string values under a key) and lists (ordered collections of strings under a key). Backed by https://api.quicknode.com/kv/rest/v1/.

Stores a single string value under a key.

Parameters: key (string, required), value (string, required).

Returns: nothing.

# Python
await qn.kvstore.create_set(key="my-key", value="hello")

Paginated page of key/value entries.

Parameters (all optional): limit (i64), cursor (string).

Returns: GetSetsResponse — { data: KvSetEntry[], cursor: string }.

# Python
resp = await qn.kvstore.get_sets()

Returns the value stored under a key.

Parameters: key (string, required).

Returns: GetSetResponse with value.

# Python
resp = await qn.kvstore.get_set("my-key")

Adds and/or deletes multiple sets in a single request.

Parameters (at least one required): add_sets (map<string,string>, optional), delete_sets (string[], optional).

Returns: nothing.

# Python
await qn.kvstore.bulk_sets(
    add_sets={"k1": "v1"},
    delete_sets=["old-key"],
)

Deletes a single set.

Parameters: key (string, required).

Returns: nothing.

# Python
await qn.kvstore.delete_set("my-key")

Creates a list under a key, seeded with the initial items.

Parameters: key (string, required), items (string[], required).

Returns: nothing.

# Python
await qn.kvstore.create_list(key="my-list", items=["0xabc", "0xdef"])

Paginated page of list keys.

Parameters (all optional): limit (i64), cursor (string).

Returns: GetListsResponse — { data: { keys: string[] }, cursor: string }.

# Python
resp = await qn.kvstore.get_lists()

Paginated page of items for a specific list.

Parameters: key (string, required); optional limit (i64), cursor (string).

Returns: GetListResponse — { data: { items: string[] }, cursor: string }.

# Python
resp = await qn.kvstore.get_list("my-list")

Adds and/or removes items in a single operation.

Parameters: key (string, required); optional: add_items (string[]), remove_items (string[]).

Returns: nothing.

# Python
await qn.kvstore.update_list(
    "my-list",
    add_items=["0x456"],
    remove_items=["0xabc"],
)

Appends a single item to a list.

Parameters: key (string, required), item (string, required).

Returns: nothing.

# Python
await qn.kvstore.add_list_item("my-list", "0x123")

Checks whether a list contains a specific item.

Parameters: key (string, required), item (string, required).

Returns: ListContainsItemResponse with exists: bool.

# Python
resp = await qn.kvstore.list_contains_item("my-list", "0x123")

Removes a single item from a list.

Parameters: key (string, required), item (string, required).

Returns: nothing.

# Python
await qn.kvstore.delete_list_item("my-list", "0x123")

Deletes a list and all of its items.

Parameters: key (string, required).

Returns: nothing.

# Python
await qn.kvstore.delete_list("my-list")

Accessed as qn.sql. Runs SQL queries against indexed blockchain data and fetches the database schema. Backed by https://api.quicknode.com/sql/rest/v1/.

Executes a SQL query against a cluster and returns the result set. Paginate by writing LIMIT/OFFSET into the SQL.

Parameters: query (str, required), cluster_id (str, required).

Returns: QueryResponse — meta (list of ColumnMeta, each with name and column_type), data (list[dict], rows keyed by column name), rows, rows_before_limit_at_least, statistics (QueryStatistics with elapsed, rows_read, bytes_read), and credits.

# Python
resp = await qn.sql.query(
    query="SELECT action_type, user FROM hyperliquid_system_actions ORDER BY block_time DESC LIMIT 100",
    cluster_id="hyperliquid-core-mainnet",
)
print(resp.rows, resp.data[0])

Fetches the database schema for a cluster: table names, columns, types, sort keys, and partition strategies.

Parameters: cluster_id (str, required).

Returns: ChainSchema — chain, cluster_id, and tables (list of TableSchema, each with name, engine, total_rows, partition_key, sorting_key, and columns of ColumnSchema with name and column_type).

# Python
schema = await qn.sql.get_schema("hyperliquid-core-mainnet")
print(len(schema.tables))

Tooling Access provisions a single multichain, read-only endpoint per account and mints short-lived session JWTs. qn.rpc makes JSON-RPC calls directly against that endpoint, minting and refreshing the JWT automatically — no endpoint URL or token to manage.

Tooling Access must be enabled once (admin role + eligible plan). Status reads can return HTTP 403 when feature access is unavailable; successful authentication alone does not guarantee access. The control-plane methods live on qn.admin:

# Python
status = await qn.admin.tooling_access_status()
if not status.enabled:
    await qn.admin.enable_tooling_access()  # idempotent; admin role required

# Make on-chain calls. params defaults to []; pass a list (positional) or dict.
block_number = await qn.rpc.call("eth_blockNumber")
balance = await qn.rpc.call("eth_getBalance", ["0xabc...", "latest"])

# Multichain: select a network by its multichain_urls key. Seed the map first
# (from admin.get_endpoint_urls), then pass network=.
urls = await qn.admin.get_endpoint_urls(endpoint_id)
mc = urls.data.multichain_urls if urls.data else {}
qn.rpc.set_networks({k: v.http_url for k, v in (mc or {}).items()})
slot = await qn.rpc.call("getSlot", network="solana-mainnet")

# Custom endpoint URL: send to a fully-formed HTTP URL, bypassing Tooling Access
# and the JWT (no Authorization header). Per-call via endpoint_url=, or
# client-wide via RpcConfig(endpoint_url=...). endpoint_url and network are
# mutually exclusive (a custom URL is not multichain-routed).
block = await qn.rpc.call("eth_blockNumber", endpoint_url="https://my-endpoint.example/rpc")

# A JSON-RPC error member is raised as RpcError (with .code and .message).
from quicknode_sdk import RpcError
try:
    await qn.rpc.call("eth_getBalance", ["bad"])
except RpcError as e:
    print(e.code, e.message)

A host that persists across processes can snapshot the cached token with qn.rpc.current_token() and re-seed it via RpcConfig(seed=...) on the next construction; refresh_margin_secs (default 60) tunes how early the token is refreshed. Set RpcConfig(endpoint_url=...) to route every call to a custom HTTP URL by default (no JWT minted); a per-call endpoint_url overrides it.

Pay per RPC request with a stablecoin instead of a provisioned account + API key, against Quicknode's x402.quicknode.com and mpp.quicknode.com gateways. Configure it by setting payment on the RPC config; the SDK runs the 402 → sign → resend handshake for you. An API key is not required for this lane — build a keyless SDK.

There are four payment paths. Two pay per request; two amortize one signature over many calls.

The signer construction is derived from the scheme and pay network, never stated directly: x402/EVM signs an EIP-712 TransferWithAuthorization, x402/Solana an SPL TransferChecked in a v0 tx (the gateway sponsors gas), and MPP/Tempo a native Tempo transaction.

scheme selects the gateway for call only. The gateway_* drawdown methods always use the x402 gateway and the mpp_* channel methods always use the MPP gateway, whatever scheme is set to.

PaymentConfig fields:

network on the call is the query chain (gateway path slug), independent of the pay network. Use call_with_receipt to also get the settlement receipt (reference = settlement tx hash) — populated on the MPP lane, null/None/nil for x402.

Things to know:

• Do not log your own PaymentConfig — the key field is readable. The SDK never prints it in its own errors/Debug, but a plain print(config) will show it.
• max_amount is integer base units of the selected asset. The SDK skips any offered entry above it and refuses to sign one — a guard against an overcharging gateway.
• PaymentIndeterminateError means the paid request was sent but the response was lost. You MAY have been charged — do not blindly retry.
• x402/Solana: one payment per call. Building a payment reads the mint and a recent blockhash from a Solana RPC. The default is a public RPC that rate-limits aggressively — set svm_rpc_url to your own endpoint at any volume.

import os
from quicknode_sdk import QuicknodeSdk, SdkFullConfig, RpcConfig, PaymentConfig

qn = QuicknodeSdk(SdkFullConfig(api_key=None, rpc=RpcConfig(payment=PaymentConfig(
    scheme="x402",
    key=os.environ["QN_PAYMENT_KEY"],
    pay_network="eip155:84532",
    asset="0x036CbD53842c5426634e7929541eC2318f3dCF7e",
    max_amount="10000",
))))
resp = await qn.rpc.call_with_receipt("eth_blockNumber", [], "base-sepolia")
print(resp["result"], resp["payment_receipt"])

generate_payment_wallet("evm") creates a fresh keypair offline — no network call, no funds — for "evm", "svm", or "tempo". The private key is returned exactly once, at generation; nothing in the SDK stores or re-derives it, so persist it immediately.

from quicknode_sdk import generate_payment_wallet

wallet = generate_payment_wallet("evm")
print("fund this address:", wallet["address"])
open("payment.key", "w").write(wallet["key"])  # returned exactly once

Cheaper per call than paying per request: one SIWE or SIWS signature mints a session JWT, then each call draws a single credit from the account balance instead of signing a fresh settlement. Minting the session is free and moves no funds, so a host can re-authenticate transparently. Persist it between processes.

Fund the payment wallet out of band — the testnet faucet below, or by sending funds to payment_address() directly. Credits are provisioned against the account gateway-side.

EVM payment networks use SIWE. Solana payment networks use SIWS with an Ed25519 signature encoded as Base58.

session = await qn.rpc.gateway_authenticate()
balance = await qn.rpc.gateway_credits(session)
print("credits:", balance["credits"])
result = await qn.rpc.gateway_drawdown_call("eth_blockNumber", session, "base-sepolia")

A token_expired surfaces as an ApiError with status 401/403; re-authenticate and retry that call.

gateway_drip requests testnet tokens for the payment wallet. Circle Gateway-backed networks return transfer_id because settlement is asynchronous. Direct-transfer networks such as Arc Testnet return transaction_hash. The response is not a credit balance; call gateway_credits separately.

Open a payment channel by depositing into the escrow, then authorize each call with a cumulative voucher — one ecrecover server-side, no on-chain transaction per call.

channel = await qn.rpc.mpp_open("1000000")        # persist this dict
new_total = str(int(channel["cumulative_spent"]) + int(channel["per_call"]))
result = await qn.rpc.mpp_session_call(
    "eth_blockNumber", "base-sepolia", channel, new_total
)
# On success, store new_total as the channel's cumulative_spent.

Things to know:

• Persist the channel state. The gateway exposes no read-only channel endpoint, so a lost local record means opening (and funding) a new channel.
• mpp_status is not free. The gateway prices every session POST as a chargeable request and computes the balance from the new spend a voucher authorizes, so the probe advances cumulative_spent by per_call exactly like a call. Re-persist the advanced total. It raises PaymentUnsupportedError before any network I/O when the channel has no room left for the probe.
• The lifecycle takes no query network. A channel is scoped by the configured pay network and asset, so one channel funds calls to every supported network. Only mpp_session_call takes a network, because it routes an RPC method.
• Amounts are decimal strings, not numbers. They are u128 in the core; a Python int has no lossless u128 conversion, so pass and store them as strings.
• Advance cumulative_spent only after a success. A voucher authorizes the running total after the call; re-presenting the current high-water mark authorizes zero and is always refused with insufficient-balance.

Every binding exposes a typed exception hierarchy derived from the core SdkError enum (crates/core/src/errors.rs). Catch the base class (QuicknodeError) for any SDK-originated failure, or a specific subclass to branch on transport vs. API semantics.

Class names: Importable from quicknode_sdk: QuicknodeError, ConfigError, HttpError, TimeoutError, ConnectionError, ApiError, DecodeError, RpcError, PaymentError, PaymentUnsupportedError, PaymentRejectedError, PaymentIndeterminateError.

# Python
from quicknode_sdk import ApiError, TimeoutError
try:
    await qn.admin.show_endpoint("missing")
except ApiError as e:
    if e.status == 404:
        print(f"not found: {e.body}")
    else:
        raise
except TimeoutError:
    print("timed out")

MIT