ahstn/oceans-llm / .agents/skills/verify-oceans-admin
verify-oceans-admin
VerifiedVerify the Oceans LLM admin control plane and backend gateway against the real local stack, seeded demo data, and bounded live providers when required. Use for user-path checks of sign-in, models, API keys, observability, usage costs, spend controls, batches, agent sessions, request logs, Anthropic Messages thinking and effort compatibility, MCP registry and tool access, OpenRouter routing, guardrails, and Skills storage.
scripts/mcp-candidates.mjs
View on GitHub// Candidate file contract for `drive mcp`. Validation runs before the browser starts.
// Each candidate registers a server and makes bounded upstream calls, so the cap limits
// quota and run time. Six fits the example: one public server, GitHub, and Exa and ReUI each
// paired with a negative control.
export const MAX_CANDIDATES = 6;
const MAX_RESULT_MARKERS = 4;
export function validateCandidates(value) {
if (!Array.isArray(value?.candidates) || value.candidates.length < 1 || value.candidates.length > MAX_CANDIDATES) throw new Error(`Supply one to ${MAX_CANDIDATES} MCP candidates, including any negative-auth control`);
if (!value.candidates.some((candidate) => candidate.required !== false)) throw new Error("At least one MCP candidate must be required");
const seen = new Set();
for (const candidate of value.candidates) {
if (!/^[a-z0-9][a-z0-9-]{0,30}$/.test(candidate.key) || seen.has(candidate.key)) throw new Error("Candidate key must be unique and URL-safe");
seen.add(candidate.key);
if (typeof candidate.label !== "string" || !candidate.label.trim()) throw new Error("Candidate label is required");
const url = new URL(candidate.server_url);
if (url.protocol !== "https:" || url.username || url.password || url.hash) throw new Error("Candidate endpoint must be HTTPS without URL credentials");
if ([...url.searchParams.keys()].some((key) => /key|token|secret|auth/i.test(key))) throw new Error("Do not place credentials in the MCP URL");
if (!["none", "gateway_static_header", "gateway_bearer_token"].includes(candidate.auth_mode)) throw new Error("This proof supports public or gateway-authenticated discovery only");
const auth = candidate.auth_config ?? {};
if (Object.keys(auth).some((key) => !["secret_ref", "header_name", "value_template", "query_param"].includes(key))) throw new Error("Auth config must contain only an environment reference and an optional header name, value template, or query parameter name");
if (candidate.auth_mode !== "none" && !/^env\/OCEANS_MCP_DISCOVERY_[A-Z0-9_]+$/.test(auth.secret_ref ?? "")) throw new Error("Auth secret_ref must use env/OCEANS_MCP_DISCOVERY_*");
const control = candidate.expect_tool_error || candidate.expect_discovery_error;
if (control && candidate.auth_mode === "none") throw new Error("Negative-auth control requires an authenticated server mode");
if (candidate.expect_tool_error && candidate.expect_discovery_error) throw new Error("A control expects either a tool error or a discovery error, not both");
if (candidate.expect_result_markers !== undefined) {
const markers = candidate.expect_result_markers;
if (control) throw new Error("Result markers apply only to successful calls");
if (!Array.isArray(markers) || markers.length < 1 || markers.length > MAX_RESULT_MARKERS ||
markers.some((marker) => typeof marker !== "string" || !marker.trim() || marker.length > 120)) throw new Error(`expect_result_markers must list one to ${MAX_RESULT_MARKERS} short strings`);
}
// A discovery control never reaches a tool call, so it carries no call.
if (candidate.expect_discovery_error) {
if (candidate.call !== undefined) throw new Error("A discovery-error control makes no tool call; remove its call");
continue;
}
if (!candidate.call || typeof candidate.call.name !== "string" || !candidate.call.arguments || Array.isArray(candidate.call.arguments) || typeof candidate.call.arguments !== "object") throw new Error("Supply a reviewed read-only tool call and synthetic arguments");
}
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |