Menu
SkillsMP has collected 23 skills from OpenRouterTeam/docs. Open a skill to review its source and details.
Showing 23 of 23 collected skills.
Detect accounts reselling US frontier models through a relay or proxy, using account-level user-agent and egress-network dominance in default.generations, and file the survivors as Sentinel ban candidates. Use when asked to sweep for proxy/relay traffic,…
Implement a new batch provider (e.g. Anthropic, Gemini) from its committed research note — file-by-file recipe for the endpoint-family schema entry, skin selection/creation, adapter lifecycle methods, enum/factory registration, runtime env plumbing, and the…
Write and audit tests for the Batch API — colocated bun:test unit tests in packages/batch and services/batch-api, the vitest e2e happy path in tests/e2e/api/batches against the local Tilt batch stack + fake provider, and the test-quality audit pass. Sub-skill…
Capture real sync-API responses and provider batch output and turn them into committed JSONL fixtures for Batch API tests — batch input files, batch output lines, sync-response golden vectors for parity tests, and the Zod-validated loader pattern. Covers both…
Run and create end-to-end tests after creating a PR AND after every subsequent code push that changes behavior. For API changes, run the e2e test suite and inspect dev-fs-logs. For frontend changes, record a video, review it for bugs, fix any issues found,…
Query the ClickHouse analytics cluster for users whose current (intraday-corrected) balance is negative, using the same balance-correction and eligibility logic as the negative-balance-usage-sync cron. Use when asked who currently has a negative balance, how…
Adjudicate a Sentinel case that groups accounts by a shared mail relay or mail-domain family — decide whether the relay is operator infrastructure or a commercial host, then derive the per-account signals that justify or refuse enforcement.
Challenge the tests you just wrote with StrykerJS mutation testing. Use after adding or changing tests for a pure function, parser, serializer, or other logic-heavy module, to find assertions the suite executes but never checks. Also use it before you open a…
Use the Sentinel ban-candidates CLI to propose ban candidates and read back suggestions and their targets. Detection agents use the ingest key.
Cloudflare Worker startup CPU time limit optimization history, measurement tools, and architecture reference for cfw-api. Use when investigating startup failures, optimizing bundle size, or planning changes that affect cfw-api's startup path.
Investigate Batch API production issues using Datadog logs and metrics. Given a batch job ID, Clerk entity ID, or symptom such as "batch stuck", "results 402", "usage missing", or "submit failing", traces the job through ingress, acceptance, asynchronous…
Playbook for OPE-5618: migrating projects/web server actions and RSC loaders that call @openrouter-monorepo/db directly onto cfw-frontend-api private routes + the TanStack data-layer. Covers the backend route port (auth parity, error handling, schemas),…
Land a benchmark that already exists upstream in the bench-harness into this monorepo — subtree pull, Temporal start-request registration, benchmark OpenAPI regeneration, mission-control catalog metadata, and the subtree-integrity check. Use after the…
Seed deterministic ClickHouse usage data for a specific Clerk user and workspace so Activity Explorer verification has predictable top-N, Other, granularity, and export data.
Read-only audit of a Batch API PR or PR stack against the batch layer boundary rules, streaming/memory invariants, error-handling contracts, and stack scoping. Produces a checklist report with file:line evidence for every gap. Run before requesting review on…
Get a model + endpoint into the local Postgres DB and test it via curl against the local cfw-api. Preferred path is prod-first via the seed CSVs; manual SQL staging is the fallback for models/adapters not yet in prod.
Measurement methodology for request-correlated memory leaks in cfw Worker isolates — the floor-vs-request-count probe, control-arm subtraction, class-count structural check, incoming-edge retainer histograms, and cut simulation with a positive control. Use…
Validate Tilt service readiness before running tests. Covers unit tests (no Tilt needed), E2E / manual web / cURL tests (services must be ready), restarting and recovering resources, manual-trigger resource reference, Infisical machine-identity auth for…
Run and test the full intern stack locally — create wizard, Slack app mint, OAuth install, provisioning, and a real GCP VM whose bot answers in Slack — with no deploy. Covers the prerequisites, the exact start-up order, and the failure modes whose error…
Run a real intern locally with no provisioning: `tilt up -- --interns` boots the production ori-runtime container and a dev-only web chat talks to it. Use to test intern behaviour without GCP, Slack, OAuth, or the vault.
Weekly audit for slow PostgreSQL queries using Datadog DBM metrics, query samples, and explain plans. Produces evidence-backed optimization recommendations for openrouter-web.
Daily adoption report for rerank models. Combines ClickHouse usage metrics (default.generations, api_type='rerank') with status-field reliability signals to track daily and weekly users, requests, revenue, cost per request, model mix, and failures. Runs daily…
Daily adoption report for text-to-speech (TTS) models. Combines ClickHouse usage metrics (default.generations, api_type='tts') with cfw-tts-api Datadog reliability logs to track adoption since launch — weekly & daily unique users (new vs repeat), cost/revenue…