guidewire-sdk-patterns
Provides robust patterns for building Guidewire REST API clients that handle concurrency, throttling, and retries.
Install
mkdir -p .claude/skills/guidewire-sdk-patterns && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/3525" && unzip -o skill.zip -d .claude/skills/guidewire-sdk-patterns && rm skill.zipInstalls to .claude/skills/guidewire-sdk-patterns
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Build a production-grade Guidewire Cloud API client that survives the request-side failures — 409 checksum conflicts on PATCH/PUT, 429 quota throttling, offsetToken pagination drift, retry-unsafe POSTs, and unstructured error responses. Use when designing an HTTP client wrapper around PolicyCenter, ClaimCenter, or BillingCenter REST endpoints. Trigger with "guidewire client", "guidewire sdk", "checksum 409", "guidewire pagination", "guidewire rate limit", "Retry-After".Key capabilities
- →Handle 409 checksum conflicts for safe mutations
- →Implement `Retry-After`-aware rate limiting for 429 responses
- →Paginate through Cloud API results using `offsetToken`
- →Ensure retry-safe writes with `Idempotency-Key`
- →Provide typed error handling for Guidewire Cloud API responses
How it works
The skill provides patterns for building a production-grade Guidewire Cloud API client by implementing checksum round-trips for mutations, `Retry-After`-aware rate limiting, `offsetToken` pagination, `Idempotency-Key` for writes, and structured error handling.
Inputs & outputs
When to use guidewire-sdk-patterns
- →Implementing Guidewire REST API clients
- →Handling 409 checksum conflicts
- →Managing API rate limits
- →Building pagination-safe requests
About this skill
Guidewire SDK Patterns
Overview
Build the Cloud API request layer that runs in production. This skill assumes the auth layer from guidewire-install-auth is in place — bearer tokens come from the cached token provider, not a static API_KEY. What this layer adds: safe mutations under optimistic locking, retry-aware writes, quota-friendly request pacing, complete pagination, and a typed error surface that business logic can pattern-match.
Five production failures this skill prevents:
- 409 Conflict storms — client GETs a record, lets the user (or workflow) edit it, then PATCHes back without the latest
checksum; the next concurrent edit causes both PATCHes to lose. - 429 cascades — every retry on
429happens at the same backoff, all clients hammer the endpoint together, the tenant rate quota stays pinned. - Silent data loss in pagination — client treats
pageSize=100as "all results"; misses page 2 onward; reports inflated coverage to downstream. - Duplicate writes on retry — POST to create-claim times out at the load balancer, client retries, two claims land with the same FNOL.
- Generic error handling — every non-2xx response collapses into "request failed", losing the
userMessage,errors[].type, andattributesstructure the API actually returns.
Prerequisites
- A working auth layer per
guidewire-install-auth—getToken()returning a cached, scope-validated bearer - Cloud API endpoints reachable on
[TENANT].guidewire.net(PC/CC/BC base URLs in env) - Node 20+ or equivalent runtime supporting
fetch,AbortController, andcrypto.randomUUID - Familiarity with the Cloud API response envelope (
data[],attributes,checksum,links) — seeguidewire-install-auth/references/API_REFERENCE.md
Instructions
Implement the patterns below as composable layers on top of fetch. Each pattern targets one of the five production failures listed in Overview; do not skip any layer in production code.
1. Checksum round-trip for safe mutations
Every Cloud API resource carries a checksum. PATCH and PUT must echo the latest checksum in the request body, or the API returns 409 Conflict to protect concurrent writers. Wrap mutations in a fetch-then-mutate helper that does the round-trip automatically.
export async function patchResource<T>(path: string, mutate: (current: T) => Partial<T>): Promise<T> {
const token = await getToken();
const getRes = await fetch(`${BASE}${path}`, { headers: { Authorization: `Bearer ${token}` } });
if (!getRes.ok) throw await mapError(getRes, "GET", path);
const { data: current } = await getRes.json();
const patchRes = await fetch(`${BASE}${path}`, {
method: "PATCH",
headers: { Authorization: `Bearer ${token}`, "Content-Type": "application/json" },
body: JSON.stringify({
data: { attributes: mutate(current.attributes), checksum: current.checksum },
}),
});
if (!patchRes.ok) throw await mapError(patchRes, "PATCH", path);
return (await patchRes.json()).data;
}
When 409 still occurs (rare race), the caller should retry the helper itself, not the inner PATCH — the inner PATCH would just hit 409 again with the same stale checksum.
2. Retry-After-aware rate limiting
On 429 Too Many Requests, Cloud API returns a Retry-After header that is either an integer (seconds) or an HTTP date. Honour both forms; never use a fixed backoff. Combine with exponential-with-jitter for transient 5xx so concurrent clients do not synchronize their retries.
async function backoffFor(res: Response, attempt: number): Promise<number> {
if (res.status === 429) {
const ra = res.headers.get("Retry-After");
if (ra) return /^\d+$/.test(ra) ? Number(ra) * 1000 : Math.max(0, Date.parse(ra) - Date.now());
}
// Decorrelated jitter: caps the herd, keeps p99 sane
return Math.min(30_000, Math.random() * (200 * 2 ** attempt));
}
3. Pagination via offsetToken
Cloud API does not page by number. The response carries links.next.href until the data is exhausted; absence of links.next is the terminator. Iterate as a generator so callers do not buffer the whole dataset.
export async function* paginate<T>(path: string): AsyncGenerator<T> {
let next: string | null = path;
while (next) {
const token = await getToken();
const res = await fetch(`${BASE}${next}`, { headers: { Authorization: `Bearer ${token}` } });
if (!res.ok) throw await mapError(res, "GET", next);
const body = await res.json();
for (const item of body.data) yield item.attributes as T;
next = body.links?.next?.href ?? null;
}
}
4. Idempotency-Key for retry-safe writes
POST is not naturally idempotent. A timeout at the load balancer can cause the client to retry a request the API already processed — duplicate claim, duplicate payment, duplicate activity. Cloud API accepts an Idempotency-Key header; supplied keys deduplicate within a 24-hour window. Generate a v4 UUID per logical operation (not per HTTP attempt).
export async function createClaim(payload: NewClaim): Promise<Claim> {
const idempotencyKey = crypto.randomUUID(); // generated once, reused across retries
return retryable(async () => {
const token = await getToken();
const res = await fetch(`${BASE}/cc/rest/v1/claims`, {
method: "POST",
headers: { Authorization: `Bearer ${token}`, "Content-Type": "application/json", "Idempotency-Key": idempotencyKey },
body: JSON.stringify({ data: { attributes: payload } }),
});
if (!res.ok) throw await mapError(res, "POST", "/cc/rest/v1/claims");
return (await res.json()).data;
});
}
5. Structured error mapping
Cloud API error bodies are structured: { userMessage, errors: [{ message, type, attributes }] }. Lift the structure into a typed exception so business logic can pattern-match (if (e instanceof GwBusinessRuleError) rather than if (msg.includes("policy"))).
export class GwError extends Error {
constructor(public status: number, public op: string, public path: string,
public body: { userMessage?: string; errors?: { type?: string; message: string; attributes?: unknown }[] }) {
super(body.userMessage ?? `${op} ${path} failed: ${status}`);
}
is(type: string): boolean { return !!this.body.errors?.some(e => e.type === type); }
}
async function mapError(res: Response, op: string, path: string): Promise<GwError> {
const body = await res.json().catch(() => ({}));
return new GwError(res.status, op, path, body);
}
Output
A production-grade Cloud API client ships with all five layers wired:
patchResource(path, mutate)andputResource(path, replace)helpers performing checksum round-trip — callers never touch rawchecksumfields.- A
retryable(fn)wrapper usingbackoffFor()to honourRetry-Afterand apply jitter on transient5xx. paginate(path)async generator yielding all records without buffering; callers iterate naturally.Idempotency-Keyset on every POST, with the key generated once per logical operation and persisted across retries.GwErrorexceptions exposing.status,.op,.path, structured.body, and.is(type)for pattern matching.
Examples
Example 1 — Safe policy update under concurrent writers
const updated = await patchResource<Policy>("/pc/rest/v1/policies/pc:8001", current => ({
expirationDate: addYears(current.expirationDate, 1),
}));
The helper performs GET, applies the mutation, PATCHes back with the latest checksum. If two services update the same policy simultaneously, one wins, the other gets 409 and can retry the helper for a clean re-merge.
Example 2 — Bulk import with quota-friendly pacing
const claims: NewClaim[] = await loadFnolBatch();
for (const claim of claims) {
await retryable(() => createClaim(claim)); // honours Retry-After on 429, jitter on 5xx
}
A 10,000-record batch import that respects the tenant quota without manual rate calculations. The single-flight token cache from guidewire-install-auth keeps Hub calls bounded as well.
Example 3 — Pagination without surprises
let total = 0;
for await (const acct of paginate<Account>("/pc/rest/v1/accounts?pageSize=100")) {
await indexAccount(acct);
total++;
}
console.log(`indexed ${total} accounts`);
paginate() follows links.next.href until exhausted; the caller cannot accidentally truncate at page 1.
Error Handling
| Error | Cause | Solution |
|---|---|---|
409 Conflict on PATCH/PUT | stale checksum — another writer mutated the resource between your GET and PATCH | retry the outer helper (re-GET + re-mutate); retrying the PATCH alone re-uses the stale checksum and will 409 again |
429 Too Many Requests | tenant quota exceeded | honour Retry-After; if it recurs, the client is missing a token-bucket layer or another integration shares the tenant quota |
400 Bad Request with errors[].type === "invalid-attribute" | validation failure on a field | inspect errors[].attributes for the offending path; do not blanket-retry |
422 Unprocessable Entity with errors[].type === "rule-violation" | Gosu business rule rejected the payload | not a transport failure — surface userMessage to the caller, do not retry |
502/503/504 | transient gateway or upstream blip | retry with backoffFor() jitter, max 3 attempts |
Duplicate POST when Idempotency-Key is omitted or per-attempt | client treated each retry as a new logical operation | generate the key once per logical operation, reuse across retries |
| Truncated dataset from a paginating endpoint | caller treated first page as complete | replace ad-hoc pagination with the paginate() generator |
userMessage lost in logs | error wrapper stringified the body and discarded structure | use GwError and log .body.errors separately |
For a deeper r
Content truncated.
When not to use it
- →When a static API key is used instead of a cached token provider
- →When ad-hoc pagination is preferred over a generator-based approach
Prerequisites
Limitations
- →Requires an existing authentication layer from `guidewire-install-auth`
- →Assumes familiarity with the Cloud API response envelope
- →Does not cover the batch endpoint, `included` parameter, or HTTP/2 connection pooling
How it compares
This skill prevents common production failures in Guidewire Cloud API interactions by implementing specific patterns for optimistic locking, rate limiting, pagination, and idempotency, unlike basic HTTP client implementations.
Compared to similar skills
guidewire-sdk-patterns side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| guidewire-sdk-patterns (this skill) | 2 | 27d | Review | Advanced |
| deepgram-performance-tuning | 3 | 27d | Review | Intermediate |
| graphql | 6 | 6mo | No flags | Advanced |
| groq-performance-tuning | 1 | 27d | No flags | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by jeremylongshore
View all by jeremylongshore →You might also like
deepgram-performance-tuning
jeremylongshore
Optimize Deepgram API performance for faster transcription and lower latency. Use when improving transcription speed, reducing latency, or optimizing audio processing pipelines. Trigger with phrases like "deepgram performance", "speed up deepgram", "optimize transcription", "deepgram latency", "deepgram faster".
graphql
davila7
GraphQL gives clients exactly the data they need - no more, no less. One endpoint, typed schema, introspection. But the flexibility that makes it powerful also makes it dangerous. Without proper controls, clients can craft queries that bring down your server. This skill covers schema design, resolvers, DataLoader for N+1 prevention, federation for microservices, and client integration with Apollo/urql. Key insight: GraphQL is a contract. The schema is the API documentation. Design it carefully.
groq-performance-tuning
jeremylongshore
Optimize Groq API performance with caching, batching, and connection pooling. Use when experiencing slow API responses, implementing caching strategies, or optimizing request throughput for Groq integrations. Trigger with phrases like "groq performance", "optimize groq", "groq latency", "groq caching", "groq slow", "groq batch".
openrouter-streaming-setup
jeremylongshore
Implement streaming responses with OpenRouter. Use when building real-time chat interfaces or reducing time-to-first-token. Trigger with phrases like 'openrouter streaming', 'openrouter sse', 'stream response', 'real-time openrouter'.
perplexity-multi-env-setup
jeremylongshore
Configure Perplexity across development, staging, and production environments. Use when setting up multi-environment deployments, configuring per-environment secrets, or implementing environment-specific Perplexity configurations. Trigger with phrases like "perplexity environments", "perplexity staging", "perplexity dev prod", "perplexity environment setup", "perplexity config by env".
serving-llms-vllm
davila7
Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.