MCP server giving AI agents Lightning payments, L402 API access, trust verification, and service discovery. Free, open-source, NWC-native.
Lightning-mcp defines 11 tools with reasonable naming (verb_noun pattern) and descriptions. However, input schemas are incomplete: most parameters lack formal type definitions beyond Zod declarations in code. Descriptions are present but generic (avg ~100 chars, below the 194-char baseline). Error handling returns plain text rather than structured guidance. No output schemas documented. Tool composition is sound (each does one thing), but parameter descriptions lack constraint details (ranges, formats, enums). Security is reasonable (no secrets in params), but no permission gates or audit logging visible.
Fetch a URL that may be behind an L402 (Lightning-gated) paywall. Automatically pays the invoice and retries with proof of payment. Returns the response data.
Check the current balance of the connected Lightning wallet (in sats).
Query trust attestations for a Nostr pubkey and calculate a trust score based on service quality, general trust, and temporal decay.
Create a Lightning invoice to receive a payment.
Decode a BOLT11 invoice to inspect its amount, description, expiry, and payment hash.
Discover AI agents and services registered in the agent-discovery directory. Optionally filter by tag.
Input schemas lack formal JSON Schema type definitions. Zod validators exist in code but are not exposed as structured inputSchema objects in tool registration. Parameters like 'invoice', 'address', 'amountSats' have descriptions but no explicit type, minimum, maximum, or pattern constraints in the schema.
No output schemas documented. Tools return JSON text responses but do not declare the structure (fields, types, required keys) of the returned object. LLMs cannot plan downstream tool calls or extract fields reliably.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2026-07-28+ | v2 |
Pay multiple Lightning addresses in a batch. Resolves each address and pays it.
Pay multiple Lightning invoices in a batch. Returns results for each payment.
Pay a BOLT11 Lightning invoice. Returns the payment preimage on success.
Pay a Lightning address ([email protected] format). Resolves the address, creates an invoice, and pays it.
Resolve a Lightning address to its LNURL endpoint and metadata.
Error responses are plain text strings without structured guidance. Errors like 'Payment failed: <message>' do not tell the LLM whether to retry, ask the user, or abort. No error classification (retryable vs fatal) or recovery hints.
Irreversible operations (pay_invoice, pay_lightning_address, pay_batch, pay_addresses, access_l402) lack confirmation or dry-run support. Agents can trigger payments without a confirmation step, risking accidental fund loss.
Parameter descriptions lack constraint details. E.g., 'amountSats' has no minimum/maximum; 'maxSats' defaults are mentioned in text but not enforced in schema; 'tag' in discover_services has no enum of valid tags. Constraints should be formal (enum, pattern, min/max) not prose.