Index ranks. Resolve routes. Rhumb is an agent gateway for external tools: Index scores and compares services; Resolve routes supported capability calls through governed execution rails with receipts.
🌐 rhumb.dev · ⚡ Quickstart · 🧭 Resolve · 💵 Pricing · 📊 Leaderboard · 📖 Methodology · 🔑 Trust
For agents: See
llms.txtfor machine-readable documentation andagent-capabilities.jsonfor structured capability metadata.
npx rhumb-mcp@latestZero config. Discovery tools work immediately — no signup, no governed API key.
For execution, pass your governed API key:
RHUMB_API_KEY=your_key npx rhumb-mcp@latestcurl "https://api.rhumb.dev/v1/services/stripe/score"
# See supported execution routes before you execute
curl "https://api.rhumb.dev/v1/capabilities/email.send/resolve"All read endpoints are public, including Resolve readiness checks. Estimate and execute require an authenticated payment path.
Use Rhumb Index when an agent needs to discover and evaluate services. Use Rhumb Resolve when the task is on a supported capability path and you want governed execution with an explicit receipt.
Agents need external tools. Choosing the right one is hard — not because of feature lists, but because of:
- auth and signup friction
- provisioning reality vs. marketing claims
- schema instability
- failure recovery when no human is watching
- hidden costs and rate limits
Rhumb makes those constraints visible before you commit.
Rhumb is strongest today for research, extraction, generation, and narrow enrichment.
Treat broader multi-system business automation as future scope, not the current launch promise. Use Layer 2 capabilities for real work now, and treat Layer 3 as beta with an intentionally sparse public catalog.
999 scored services across 50+ domains. Each gets an AN Score (0–10) measuring execution quality, access readiness, and agent autonomy support.
find_services— Search indexed Services by what you need them to doget_score— Get the full AN Score breakdown for a Service: execution quality, access readiness, autonomy level, tier label, and freshnessget_alternatives— Find alternative Services, ranked by AN Scoreget_failure_modes— Get known failure patterns, impact severity, and workarounds for a servicediscover_capabilities— Browse Capabilities by domain or search textresolve_capability— Given a Capability ID, and optionally a credential mode, returns ranked providers with health status, cost per call, auth methods, endpoint patterns, execute guidance, and machine-readable recovery fields like recovery_hint.resolve_url, recovery_hint.credential_modes_url, and, when applicable, recovery_hint.alternate_execute_hint or recovery_hint.setup_handoff, plus typo recovery when the capability ID is wrong
Discovery breadth is wider than current execution coverage. The index is broader than what Rhumb can execute today.
435 capability definitions across 28 callable providers today. Resolve chooses the best-fit supported provider for the call using AN Score, availability / circuit state, estimated cost, latency proxy, credential mode, and explicit policy constraints.
execute_capability— Call a Capability through Rhumb Resolveresolve_capability— Given a Capability ID, and optionally a credential mode, returns ranked providers with health status, cost per call, auth methods, endpoint patterns, execute guidance, and machine-readable recovery fields like recovery_hint.resolve_url, recovery_hint.credential_modes_url, and, when applicable, recovery_hint.alternate_execute_hint or recovery_hint.setup_handoff, plus typo recovery when the capability ID is wrongestimate_capability— Estimate the active execution rail, cost, and health before a Capability call; anonymous direct system-of-record paths also preserve machine-readable execute_readiness handoffsget_receipt— Retrieve an execution receipt by ID- Budget enforcement, credential management, and execution telemetry included
Best current fit: research, extraction, generation, and narrow enrichment. Treat general business-agent automation and broad multi-system orchestration as future scope, not the current launch promise.
| Surface | What it is for | Current honest boundary |
|---|---|---|
| Rhumb Index | Free service discovery, AN Score lookup, alternatives, and failure-mode research | Broad discovery is not the same as execution readiness |
| Rhumb Resolve | Governed execution for supported capabilities with estimates, receipts, budgets, and telemetry | 28 callable providers today; best fit is research, extraction, generation, and narrow enrichment |
| MCP package | Agent-native entry point for Claude, Cursor, and other MCP clients | Discovery works without auth; execution needs a governed key, wallet-prefund, or x402 where supported |
| API | Public read endpoints plus authenticated execution endpoints | Use current API responses as source of truth for readiness and callable coverage |
| Path | How it works |
|---|---|
| Rhumb-managed | Rhumb holds the credential — zero setup for the agent |
| BYOK | Bring your own provider API key. Rhumb routes, you authenticate |
| Agent Vault | Your key, encrypted and stored — Rhumb injects at call time |
- Governed API key — sign up, get a key, prepaid credits
- x402 / USDC — no signup, pay per call on-chain
- Service = vendor Rhumb evaluates and compares
- Capability = executable action like
email.send - Recipe = deterministic multi-step workflow on top of capabilities (beta, sparse public catalog)
- Layer 2 is the default path — start with governed API key or wallet-prefund on
X-Rhumb-Key, discover a Service, choose a Capability, estimate, then execute - Start with managed superpowers first — bring BYOK or Agent Vault only when the workflow touches your own systems
- Default auth for repeat traffic = governed API key or wallet-prefund on
X-Rhumb-Key - Bring BYOK or Agent Vault only when provider control is the point
- Use x402 when zero-signup per-call payment matters more than repeat throughput
Canonical onboarding map: https://rhumb.dev/docs#resolve-mental-model
rhumb-mcp exposes 21 tools:
Discovery
find_services— Search indexed Services by what you need them to doget_score— Get the full AN Score breakdown for a Service: execution quality, access readiness, autonomy level, tier label, and freshnessget_alternatives— Find alternative Services, ranked by AN Scoreget_failure_modes— Get known failure patterns, impact severity, and workarounds for a servicediscover_capabilities— Browse Capabilities by domain or search textresolve_capability— Given a Capability ID, and optionally a credential mode, returns ranked providers with health status, cost per call, auth methods, endpoint patterns, execute guidance, and machine-readable recovery fields like recovery_hint.resolve_url, recovery_hint.credential_modes_url, and, when applicable, recovery_hint.alternate_execute_hint or recovery_hint.setup_handoff, plus typo recovery when the capability ID is wrong
Execution
execute_capability— Call a Capability through Rhumb Resolveestimate_capability— Estimate the active execution rail, cost, and health before a Capability call; anonymous direct system-of-record paths also preserve machine-readable execute_readiness handoffscredential_ceremony— Get step-by-step instructions to obtain API credentials for a Servicecheck_credentials— Inspect live credential-mode readiness, globally or for a specific Capabilityrhumb_list_recipes— List the current published Rhumb Layer 3 recipe catalogrhumb_get_recipe— Get the full published definition for a Rhumb recipe, including input/output schemas and step topologyrhumb_recipe_execute— Execute a published Rhumb Layer 3 recipe once one is live in the public catalogget_receipt— Retrieve an execution receipt by ID
Billing
budget— Check or set your call spending limitspend— Get your spending breakdown for a billing period: total USD spent, call count, average cost per call, broken down by Capability and by providercheck_balance— Check your current Rhumb credit balance in USDget_payment_url— Get a checkout URL to add credits to your Rhumb balanceget_ledger— Get your billing history: charges (debits), top-ups (credits), and auto-reload events
Operations
routing— Get or set how Rhumb auto-selects providers when you don't specify one in execute_capabilityusage_telemetry— Get your execution analytics — calls, latency, errors, costs, and provider health for your Rhumb usage
Discovery spans 999 scored services, but current governed execution spans 28 callable providers.
Note: Layer 3 recipe tooling is live, but the public catalog can still be empty. Use
rhumb_list_recipesor visit/recipesbefore assuming a workflow exists.
Best current fit: research, extraction, generation, and narrow enrichment. Treat general business-agent automation as future scope, not the current launch promise.
Base URL: https://api.rhumb.dev/v1
| Endpoint | Auth | Purpose |
|---|---|---|
GET /services/{slug}/score |
No | Score breakdown |
GET /services/{slug} |
No | Service profile + metadata |
GET /services/{slug}/failures |
No | Known failure modes |
GET /search?q=... |
No | Search services |
GET /leaderboard/{category} |
No | Category rankings |
GET /capabilities |
No | Capability registry |
GET /capabilities/{id}/resolve |
No | Ranked providers + explicit recovery_hint.* fields |
POST /capabilities/{id}/execute |
Yes | Execute a capability |
GET /capabilities/{id}/execute/estimate |
Yes | Cost estimate |
GET /telemetry/provider-health |
No | Provider health status |
GET /telemetry/usage |
Yes | Your usage analytics |
GET /pricing |
No | Machine-readable pricing |
See examples/ for runnable scripts:
| Example | What it shows | Auth needed? |
|---|---|---|
| discover-and-evaluate.py | Search → Score → Failure modes | No |
| resolve-and-execute.py | Resolve → machine-readable recovery handoff → Estimate → Execute | No for resolve, yes for estimate/execute |
| budget-aware-routing.py | Budget + cost-optimal routing | Yes |
| dogfood-telemetry-loop.py | Repeatable Resolve → telemetry verification loop | Yes |
| mcp-quickstart.md | MCP setup for Claude, Cursor, etc. | Optional |
# Try discovery right now (no auth needed)
pip install httpx && python examples/discover-and-evaluate.py
# Try the resolve walkthrough right now (no auth needed for resolve)
python examples/resolve-and-execute.pyresolve-and-execute.py will still show the ranked providers plus any machine-readable recovery handoff Rhumb already identified. Set RHUMB_API_KEY only when you want to continue into estimate and execute.
- Agent Accessibility Guidelines — making web interfaces usable by AI agents
- AN Score Methodology — scoring dimensions, weights, and rubrics
- Architecture — scoring engine design
- API Reference — endpoint details
- Repo Boundary — what stays public here vs. what lives in the private ops workspace
- Security Policy — vulnerability reporting and security architecture
rhumb/
├── packages/
│ ├── api/ # Python API (Railway)
│ ├── astro-web/ # Public website (Vercel)
│ ├── mcp/ # MCP server (npm)
│ ├── cli/ # CLI tooling
│ └── shared/ # Shared types/constants
├── examples/ # Runnable examples
├── docs/ # Public documentation only
├── scripts/ # Product tooling + verification scripts
├── artifacts/ # Curated public datasets only (raw proof outputs stay local/private)
├── llms.txt # Machine-readable docs for agents
└── agent-capabilities.json # Structured capability manifest
# API
cd packages/api && pip install -r requirements.txt && uvicorn app:app --reload
# MCP
cd packages/mcp && npm ci && npm run dev
# Web
cd packages/astro-web && npm ci && npm run devNode 24+ recommended (.nvmrc included).
Every score is disputable. If you believe a score is inaccurate:
- Read the public provider guide at rhumb.dev/providers
- Open the score-dispute GitHub template with evidence
- Or email providers@supertrained.ai for a private path
We target an initial response within 5 business days. Negative findings remain visible. Rhumb does not accept payment to change scores.
- Website: rhumb.dev
- npm: rhumb-mcp
- MCP Registry: Rhumb on MCP Registry
- X: @pedrorhumb