Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Descripción del creador (original en inglés)
MCP auth gateway: verify agent identity, enforce per-tool permissions, block replays, emit receipts.
Estos tres nunca se combinan en una sola puntuación: las opiniones, las mediciones y la popularidad son cosas distintas.
Todavía no escribimos un resumen sencillo para este. Abajo están los datos que leímos del registro.
Elige la app de IA que usas y copia la configuración.
Necesitas
¿Qué app usas?
Elige tu app arriba. La recordamos, así que la próxima vez el botón dirá “Instalar en …”.
Abre Claude Desktop → Settings → Developer → Edit Config y pega esto en claude_desktop_config.json
{
"mcpServers": {
"gateway": {
"command": "npx",
"args": [
"-y",
"@bolyra/gateway"
]
}
}
}Ejecuta esto en tu terminal
claude mcp add gateway -- npx -y @bolyra/gatewayPégalo en .cursor/mcp.json de tu proyecto, o en ~/.cursor/mcp.json
{
"mcpServers": {
"gateway": {
"command": "npx",
"args": [
"-y",
"@bolyra/gateway"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Pégalo en .vscode/mcp.json de tu proyecto
{
"servers": {
"gateway": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@bolyra/gateway"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Este se ejecuta en tu computadora, y ChatGPT solo puede conectarse a servidores en internet. Elige otra app.
Tu elección se guarda en una cookie en este dispositivo; no necesitas cuenta.
Esta configuración se genera a partir de la ficha del creador en el registro. Si el README dice otra cosa, sigue el README.
Se ejecuta en tu computadora; la lista de herramientas aparece en tu app de IA después de instalarlo.
Revisamos información que cualquiera puede ver; no hemos leído el código. Las revisiones que no aplican no se cuentan. · Revisado el 7 nov 2026
El nombre en el registro se verificó con la cuenta de GitHub bolyra, dueña del repo
Todavía no hemos leído este repo en GitHub
Todavía no hemos leído este repo en GitHub
Este se ejecuta en tu computadora; no hay endpoint en internet
Se ejecuta en tu computadora y usa ajustes locales en su lugar
Se ejecuta en tu computadora; solo podemos leer listas de herramientas de endpoints en internet
Comparado con todos los servidores activos de esta categoría en el registro. Las estrellas solo cuentan cuando quien lo publica es dueño del repo.
Superó 1 de 1 revisiones básicas aplicables. No hemos leído el código. Antes de agregar una clave o un token, revisa qué acceso pide el servidor y usa una clave que tenga solo los permisos que necesitas.
El registro no indica ninguna clave obligatoria, aunque algunos servidores te piden iniciar sesión la primera vez que los usas.
Se ejecuta en tu computadora. Tu app de IA lo inicia cuando hace falta, y los datos no pasan por los servidores del creador (a menos que el propio servidor llame a un servicio en línea).
El repo de GitHub cambió por última vez –, y la última versión (0.2.1) llegó al registro hace 5 meses.
Las apps compatibles con MCP, como Claude Desktop, Claude Code, Cursor y VS Code. También necesitas Node.js en tu computadora.
En la misma categoría: Spec Driven Development, Remote Desktop Commander, FlyDocs
Servidores de la misma categoría hechos por otros creadores, ordenados por estrellas.
Cómo leer una tarjeta:★ Estrellas de GitHubiTres pilares: reseñas · revisiones · usoiActualizado = fecha en que se publicó la última versión en el registroi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Spec-driven development workflow: your agent works issues, sessions and the status lifecycle.
Copiado tal cual del repo del creador. PopNx MCP no lo ha revisado ni lo respalda. Se quitaron HTML, imágenes y scripts. Los enlaces llevan fuera de este sitio.
Bolyra MCP Auth Gateway -- standalone reverse proxy that verifies agent ZKP credentials before forwarding requests to upstream MCP servers.
Put @bolyra/gateway in front of your MCP server; it verifies agent authority, prevents replay, enforces delegated scopes, and gives you signed audit logs for every tool call.
# Install
npm install @bolyra/gateway
cat > gateway.yaml << 'EOF' target: http://localhost:3000/mcp port: 4100 devMode: true EOF
npx @bolyra/gateway --config ./gateway.yaml
Or run directly with CLI flags:
npx @bolyra/gateway --target http://localhost:3000/mcp --dev --port 4100
+----------------------------------+
| @bolyra/gateway | Agent -------->| |--------> Upstream MCP Server (with proof | 1. Parse Authorization header | (unmodified) bundle) | 2. verifyBundle() from @bolyra/mcp| | 3. checkToolPolicy() | | 4. Nonce replay check | | 5. Emit signed receipt | | 6. Proxy request + X-Bolyra-* hdrs| | | | Config: gateway.yaml | | Receipts: ./receipts/ or stdout | +----------------------------------+
Request routing:
GET /healthz -- returns gateway status (intercepted, not proxied)tools/call -- forwarded to upstream without authtools/call -- auth verified, then forwarded if authorizednpx @bolyra/gateway [options]
Options: --target Upstream MCP server URL (required unless in config) --port Gateway listen port (default: 4100) --config Path to gateway config file (default: ./gateway.yaml) --dev Enable Bolyra Core mode (classical checks: tool policy, nonce replay, signed receipts, credential binding against registered credentials; ZK proof verification off. Without registered credentials, permission claims are self-asserted — see Credential Binding below) --credentials Credentials file (YAML/JSON map: commitment -> { permissionBitmask, expiryTimestamp? }); overrides the config's credentials section --receipt-dir Directory for receipt JSON files (default: ./receipts/) --receipt-stdout Write receipts to stdout (NDJSON) --no-receipts Disable receipt generation --network Bolyra network (default: base-sepolia) --help Show help --version Show version
CLI flags override config file values.
# gateway.yaml
target: http://localhost:3000/mcp port: 4100 network: base-sepolia
devMode: false
credentials: type: static map: "48201394857102938471029384": # credential commitment (decimal) permissionBitmask: 3 # READ_DATA | WRITE_DATA expiryTimestamp: "1893456000" # unix seconds; optional in Core mode
tools: write_file: requireBitmask: 2 # WRITE_DATA (0b10) delete_file: requireBitmask: 6 # WRITE_DATA + FINANCIAL_SMALL (0b110) maxChainDepth: 0 # Direct credentials only transfer_funds: requireBitmask: 28 # FINANCIAL_* (0b11100) minScore: 90
nonce: store: memory # "memory" (default) or "redis" maxProofAge: 300 # seconds
receipts: enabled: true issuer: "my-gateway" keyId: "k1" privateKey: "${BOLYRA_RECEIPT_KEY}" # 32-byte hex secp256k1 key, env var substitution output: file # "file", "stdout", or "webhook" dir: ./receipts/
health: enabled: true path: /healthz
Environment variables are substituted in string values using ${VAR_NAME} syntax.
For embedding the gateway middleware in your own server:
import {
createGatewayMiddleware, createGatewayProxy, loadConfig, injectBolyraHeaders, computeHmac, verifyHmac, createReceiptWriter, createHealthHandler, } from '@bolyra/gateway'; import type { GatewayConfig, GatewayMiddlewareOptions, GatewayRequest, } from '@bolyra/gateway';
// Option 1: Full proxy server const config = loadConfig({ target: 'http://localhost:3000/mcp', dev: true }); const server = createGatewayProxy({ config }); server.listen(4100);
// Option 2: Just the middleware (for custom servers) const middleware = createGatewayMiddleware({ config }); // Use in your request handler: // const authorized = await middleware(req, res, toolName);
| Export | Description |
|---|---|
createGatewayProxy(options) | Create full reverse proxy HTTP server |
createGatewayMiddleware(options) | Auth verification middleware only |
loadConfig(flags?) | Load and validate gateway configuration |
injectBolyraHeaders(authCtx, receiptId?) | Build X-Bolyra-* headers |
computeHmac(headers, secret) | HMAC-SHA256 sign X-Bolyra-* headers |
verifyHmac(headers, secret, hmac) | Verify HMAC on X-Bolyra-* headers |
createReceiptWriter(config) | Create pluggable receipt output |
createGatewayReceiptSigner(config, override?) | Resolve the ES256K receipt signer (configured key or ephemeral) |
createHealthHandler(config) | Create /healthz endpoint handler |
extractToolName(body) | Extract tool name from JSON-RPC body |
loadCredentialsFile(path) | Load a --credentials file (bare map or full section) |
buildCredentialRegistry(credentials) | Compile a static credentials section into a lookup registry |
checkCredentialBinding(bundle, authCtx, registry) | Core-mode registered-credential check (see Credential Binding) |
createStaticCredentialResolver(credentials) | Static resolveCredential for production scopeCommitment binding |
hasStaticCredentials(credentials) | True when a usable static credential map is configured |
RedisNonceStore | Redis-backed NonceStore for multi-instance deployments |
For multi-instance deployments, use Redis for shared nonce replay protection:
# gateway.yaml
nonce: store: redis maxProofAge: 300 redis: url: ${REDIS_URL} keyPrefix: "bolyra:nonce:" # optional
Library usage:
import { createGatewayMiddleware, RedisNonceStore } from '@bolyra/gateway';
const store = new RedisNonceStore({ url: 'redis://localhost:6379' }); const middleware = createGatewayMiddleware({ config, nonceStore: store });
// Graceful shutdown process.on('SIGTERM', () => store.close());
Fail-closed behavior: If Redis is unreachable, the gateway returns HTTP 500, never passes requests through. Monitor Redis availability as a critical dependency.
When a request passes verification, the gateway injects these headers into the proxied request:
| Header | Value |
|---|---|
X-Bolyra-Verified | true |
X-Bolyra-DID | did:bolyra:{network}:{commitment} |
X-Bolyra-Score | Verification score (0-100) |
X-Bolyra-Permissions | Effective permission bitmask (decimal) |
X-Bolyra-Chain-Depth | Delegation chain depth (0 = direct) |
X-Bolyra-Receipt-ID | Signed receipt id for this decision (when receipts enabled) |
X-Bolyra-HMAC | HMAC-SHA256 signature (when HMAC configured) |
The upstream server can use these headers for logging/auditing without importing Bolyra.
Every tools/call decision — allow and deny, dev mode and production — produces an ES256K-signed receipt (SignedReceipt from @bolyra/receipts: secp256k1 + keccak256, Ethereum-compatible recovery). This includes rejections with no usable proof bundle at all: a missing or malformed Authorization header still leaves a signed anonymous deny receipt, so the audit trail has no unsigned gaps.
Each receipt carries the verdict (decision.allowed), a human-readable reasonCode (which tool, which permissions were required vs. held), the agent DID when known, score, permission bitmask, delegation chain depth, hashes of the presented proof material, and the decision timestamp. Any edit to a receipt breaks its signature.
Signing key resolution:
receiptSigner option (library embedders)receipts.privateKey from the config (recommended for production)file output mode, persists it to {receipt-dir}/signer.json so auditors can pin the trust anchor.Independent verification needs only the receipt (plus, optionally, the pinned signer address) — no gateway, no database:
import { verifyReceipt } from '@bolyra/receipts';
const receipt = JSON.parse(fs.readFileSync('receipts/2026-07-10/....json', 'utf8')); verifyReceipt(receipt); // true — signature + payload hash check verifyReceipt(receipt, pinnedSigner); // true — also pins the signer address
Signatures make each receipt tamper-evident; hash-chaining makes the receipt stream tamper-evident. Every receipt the gateway signs carries chain: { seq, prevReceiptHash } inside the signed payload (seq is 0-based per gateway process; the first receipt links to the 32-zero-byte genesis sentinel) plus a top-level receiptHash that the next receipt links to. Deleting, reordering, or inserting receipts in a collected log breaks chain verification even though every remaining signature stays individually valid:
# stdout mode: collect NDJSON to a file, verify signatures + chain
bolyra receipt verify-chain receipts.ndjson --signer 0xPinnedSigner
Notes for auditors:
chain-restart.Solo se muestra el principio; lee el resto en el repo