菜单
Use when adding or changing an external HTTP provider call in SysNDD (gnomAD, Ensembl, UniProt, AlphaFold, MGI, RGD, PubMed, PubTator, MONDO, JAX/HPO) or touching external-call timeouts, retries, caching, or per-request time budgets
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Use when wiring a SysNDD Vue 3 view or component to the backend API, adding a typed API client, handling API errors, or binding backend data into BootstrapVueNext tables and tooltips — especially around Plumber response shapes and BVN table/tooltip quirks
Use when adding or changing a SysNDD DB migration, editing a SQL view (especially ndd_entity_view or the core read views), touching the migration runner or manifest, restoring a database, or diagnosing a startup failure related to migrations
Use when reviewing or writing SysNDD code for security vulnerabilities or correctness bugs — authorization/role gates, SQL/expression injection, credential and secret handling, data exposure through public or MCP surfaces, resource-exhaustion/DoS from external calls, error/info leakage, and the repo's known R/Plumber footguns — especially before merging a diff or PR
Use when changing gene clustering (STRING/Leiden functional axis or phenotype MCA/HCPC), the analysis snapshot builder or validator, the memoised clustering cache, cache coherence, or LLM cluster summaries — or deploying a clustering change so public endpoints and summaries reflect it
Use when writing, running, or fixing SysNDD R/Plumber API tests (testthat) — including database-writing tests, running a single test inside the API container, mocking external providers, or diagnosing SKIP-vs-PASS, helper-loading, or path-resolution issues
Use when adding or changing durable async jobs, worker-executed code, job handlers, queue lanes or priorities in SysNDD — or diagnosing why a submitted job never runs, stays queued, or fails with "No durable async job handler registered"
| name | sysndd-external-proxy |
| description | Use when adding or changing an external HTTP provider call in SysNDD (gnomAD, Ensembl, UniProt, AlphaFold, MGI, RGD, PubMed, PubTator, MONDO, JAX/HPO) or touching external-call timeouts, retries, caching, or per-request time budgets |
Use this skill before adding or editing any outbound HTTP call to a gene/literature/ontology provider. The rules exist so one slow upstream cannot occupy a worker for tens of seconds or poison a long-lived cache. All helpers live in api/functions/external-proxy-functions.R.
Derive the timeout/retry window from external_proxy_budget(api_name, default_timeout =, default_max =, default_tries =) or route the whole call through make_external_request(url, api_name, throttle_config, ...). Never a hardcoded req_timeout(<n>) / max_seconds = <n> literal — this is enforced by test-unit-external-budget-guard.R (a bypass was reintroduced once via GeneReviews in #389, which is why the guard exists).
Wrap fetchers in memoise_external_success_only(f, cache, source = "<provider>"), not raw memoise::memoise(). It caches successful and true not-found responses but not transient list(error = TRUE, ...) failures, so a blip does not poison the 7/14/30-day caches. Passing source = also emits one structured timing log per call:
[external-proxy] ... event=complete status=... elapsed_ms=... cache=hit|miss
gnomad/ensembl/uniprot/alphafold use the memoise source; mgi/rgd instead log via the inline external_proxy_with_timing() wrapper and omit source to avoid double-logging.
EXTERNAL_PROXY_REQUEST_MAX_SECONDS (default 15s) is wired into both universal wrappers. Once a request's accumulated external time crosses it, further external calls short-circuit to a degraded 503 (request_budget_exceeded = TRUE) without contacting the upstream. The accumulator is reset per request in the preroute hook (bootstrap/mount_endpoints.R) and per job at job start (external_proxy_request_reset(), async-job-worker.R).
.pubtatornidd_reset_external_budget() in pubtator-enrichment-collector.R), or the ceiling — designed for public request paths — caps the back half of the batch.postroute hook emits [request-timing] method=… path=… status=… duration_ms=… external_ms=… slow=… (slow over API_SLOW_REQUEST_MS, default 2000), so a slow request can be attributed to external time./health, /auth, /statistics must never call an external fetcher — enforced by test-unit-cheap-route-isolation.R.
Batch jobs with their own pacing (e.g. publication_date_backfill) intentionally keep raw httr2 retry (429 handling, Retry-After) and are outside the guard scan set — do not "fix" them into external_proxy_budget(). Per-provider budgets are tunable via env (EXTERNAL_PROXY_MONDO_*, etc.).
test-unit-external-budget-guard.R, test-unit-external-proxy-budgets.R, test-unit-external-slow-provider.R, test-unit-cheap-route-isolation.R, and the router-level test-integration-slow-provider-isolation.R.