Security building blocks for modern Node.js and TypeScript services.
Zero runtime dependencies · framework-neutral core · focused security primitives
AxiomGuard (published as @axiomnode-lab/guard) is a modular security toolkit for backend services. It provides focused primitives for signed webhooks, API keys, browser request policy, idempotency, SSRF-aware outbound requests, rate limiting, secure cookies, CSRF, CORS, security headers, environment validation, secret-safe logging and repository scanning.
Use the complete package when convenience matters, or import a focused subpath when you only need one control.
- Zero npm runtime dependencies — no transitive runtime packages to pull in.
- Fail-closed security controls — configuration errors are rejected during setup, while malformed request metadata is handled defensively by the request adapters.
- Framework-neutral — first-class adapters for Express, Fastify, Hono and web-standard Fetch runtimes.
- Provider-aware — GitHub, Stripe, Slack, Meta and Standard Webhooks out of the box, with replay protection.
- Typed end to end — from
requireEnvschemas to stable error codes.
See how it compares to helmet, cors, express-rate-limit and friends.
Requirements: Node.js 20 or newer (22 or 24 LTS recommended). AxiomGuard is published as an ES module; require() works on Node 20.19+/22.12+ through require(esm). TypeScript 5.x with moduleResolution: node16 | nodenext | bundler.
npm install @axiomnode-lab/guardAlso works with other package managers:
pnpm add @axiomnode-lab/guard
# or
yarn add @axiomnode-lab/guardIf you already use Express, the adapter is the fastest way to add defensive response headers, CORS handling and an opt-in browser request policy.
import express from 'express';
import { createExpressSecurityMiddleware } from '@axiomnode-lab/guard/adapters/express';
const app = express();
app.use(createExpressSecurityMiddleware({
cors: {
origins: ['https://app.example.com'],
allowCredentials: true,
allowMethods: ['GET', 'POST', 'PATCH'],
},
requestPolicy: {
allowedOrigins: ['https://app.example.com'],
},
}));
app.get('/health', (_req, res) => {
res.json({ ok: true });
});
app.post('/profile', (_req, res) => {
res.json({ updated: true });
});
app.listen(3000);With that configuration, ordinary requests receive AxiomGuard's defensive headers, allowed browser origins receive CORS headers, preflight requests are handled automatically, and unsafe browser requests from untrusted origins are rejected.
requestPolicy is a CSRF-oriented browser control. It does not replace authentication or authorization.
import Fastify from 'fastify';
import { createFastifySecurityHook } from '@axiomnode-lab/guard/adapters/fastify';
const app = Fastify();
app.addHook('onRequest', createFastifySecurityHook({
cors: {
origins: ['https://app.example.com'],
allowMethods: ['GET', 'POST'],
},
requestPolicy: {
allowedOrigins: ['https://app.example.com'],
},
}));import { Hono } from 'hono';
import { createHonoSecurityMiddleware } from '@axiomnode-lab/guard/adapters/hono';
const app = new Hono();
app.use('*', createHonoSecurityMiddleware({
cors: {
origins: ['https://app.example.com'],
allowMethods: ['GET', 'POST'],
},
requestPolicy: {
allowedOrigins: ['https://app.example.com'],
},
}));import { createFetchSecurityHandler } from '@axiomnode-lab/guard/adapters/fetch';
const guard = createFetchSecurityHandler({
cors: { origins: ['https://app.example.com'] },
requestPolicy: { allowedOrigins: ['https://app.example.com'] },
});
export default {
fetch: (request: Request) => guard(request, (req) => app.handle(req)),
};All adapters share one validated pipeline (createSecurityCore): configuration errors throw during setup, malformed request metadata is handled defensively, Vary is merged with existing values, and X-Powered-By is removed. Pass headers as a function to inject a per-request CSP nonce.
See Framework adapters for adapter options and Redis-backed integrations.
| Goal | Import | Start with |
|---|---|---|
| Generate and verify API keys | @axiomnode-lab/guard/api-keys |
createApiKey, verifyApiKey |
| Verify signed webhooks | @axiomnode-lab/guard/webhooks |
verifyGitHubWebhookDelivery, verifyStripeWebhook, verifySlackWebhook, verifyMetaWebhook, verifyStandardWebhook |
| Protect unsafe browser requests | @axiomnode-lab/guard/request-policy |
evaluateRequestPolicy |
| Add idempotency claims | @axiomnode-lab/guard/idempotency |
createIdempotencyFingerprint, claimIdempotencyKey |
| Guard outbound URLs and fetches | @axiomnode-lab/guard/fetch |
safeFetch |
| Rate-limit requests | @axiomnode-lab/guard/rate-limit |
checkRateLimit, createRateLimitHeaders |
| Build CORS policy | @axiomnode-lab/guard/cors |
createCorsHeaders |
| Set defensive response headers | @axiomnode-lab/guard/headers |
createSecurityHeaders |
| Create secure cookies | @axiomnode-lab/guard/cookies |
serializeCookie |
| Create and verify CSRF tokens | @axiomnode-lab/guard/csrf |
createCsrfToken, verifyCsrfToken |
| Validate environment variables (typed) | @axiomnode-lab/guard/env |
requireEnv |
| Sign or verify any HMAC payload | @axiomnode-lab/guard/crypto |
verifyHmacWebhook, signHmacWebhook, secureToken |
| Keep uploads inside a directory | @axiomnode-lab/guard/filesystem |
safePath, sanitizeFilename |
| Redact secrets and PII | @axiomnode-lab/guard/logging |
redactSecrets, maskPII |
| Scan a repository for secrets | @axiomnode-lab/guard/scanner |
scanSecrets, CLI axiomguard scan |
Generate an opaque API key, return the token once, and store only the identifier and digest.
import { createApiKey, parseApiKey, verifyApiKey } from '@axiomnode-lab/guard/api-keys';
const created = createApiKey({ prefix: 'svc' });
// Return created.token to the client once.
// Persist created.id and created.digest in your database.
// Later: find the stored digest by the public id, then verify.
const parsed = parseApiKey(presentedToken);
const record = parsed && (await db.apiKeys.findById(parsed.id));
const accepted = record !== null && verifyApiKey(presentedToken, record.digest);
if (!accepted) {
throw new Error('Invalid API key');
}Generated API keys are high-entropy random credentials. verifyApiKey() compares their SHA-256 digests in constant time; this API is not intended for human passwords. Pass { pepper } to hashApiKey/verifyApiKey to HMAC digests with a server-held secret so a leaked database cannot be verified offline.
Always verify a provider signature against the exact raw request body. Parsing and re-serializing JSON before verification can change the signed bytes.
import {
MemoryReplayStore,
verifyGitHubWebhookDelivery,
} from '@axiomnode-lab/guard/webhooks';
const replayStore = new MemoryReplayStore();
const result = await verifyGitHubWebhookDelivery(
rawBody,
request.headers['x-hub-signature-256'],
process.env.GITHUB_WEBHOOK_SECRET!,
request.headers['x-github-delivery'],
{ replayStore },
);
if (!result.ok) {
// result.reason is invalid-signature, invalid-delivery, or replay.
throw new Error(`Rejected webhook: ${result.reason}`);
}A single-process service can use MemoryReplayStore. Multi-instance deployments should use one of the Redis replay-store adapters so all instances share replay state.
Provider helpers are also available for Stripe, Slack, Meta/WhatsApp and Standard Webhooks (Svix and compatible senders):
const result = await verifyStandardWebhook(rawBody, {
id: request.headers['webhook-id'],
timestamp: request.headers['webhook-timestamp'],
signature: request.headers['webhook-signature'],
}, process.env.WEBHOOK_SECRET!, { replayStore });For providers that send a bare HMAC (hex or base64) use verifyHmacWebhook; verifyFreshHmacWebhook adds timestamp freshness and replay protection, and signedInput: 'timestamp.payload' binds the timestamp into the signature. signHmacWebhook, createStripeSignatureHeader and createSlackSignature produce signatures for tests and outbound webhooks. See API reference.
import { createCsrfToken, verifyCsrfToken } from '@axiomnode-lab/guard/csrf';
// Render the token into the form or expose it to your SPA, bound to the session.
const token = createCsrfToken(process.env.CSRF_SECRET!, { sessionId: session.id });
// On unsafe requests:
if (!verifyCsrfToken(request.body.csrf, process.env.CSRF_SECRET!, { sessionId: session.id })) {
throw new Error('Invalid CSRF token');
}Tokens are signed, expiring (2 hours by default) and bound to the session. For the signed double-submit cookie pattern pass allowUnbound: true on both sides and store the token in a __Host- cookie. Combine with requestPolicy (Fetch Metadata + Origin) for defence in depth.
Use safeFetch() when a destination can be influenced by a user or external system.
import { safeFetch } from '@axiomnode-lab/guard/fetch';
const response = await safeFetch(userSuppliedUrl, {
protocols: ['https:'],
allowedHosts: ['api.example.com', '*.cdn.example.com'],
maxRedirects: 2,
timeoutMs: 5_000,
maxResponseBytes: 5_000_000,
headers: {
accept: 'application/json',
},
});
const data = await response.json();The helper validates the initial destination and followed redirects, limits redirect depth, rejects https→http downgrades, applies one timeout across the redirect chain and the response body, caps the body size and strips credentials when a redirect crosses origins. Errors carry a stable code (SafeUrlError, SafeFetchError). Use dangerouslyAllowPrivateTargets: true only in local development.
It reduces common SSRF mistakes, but it does not replace egress controls or eliminate DNS rebinding/time-of-check-time-of-use risk. See Safe Fetch.
Bind an idempotency key to the semantics of the request that first claimed it.
import {
MemoryIdempotencyStore,
claimIdempotencyKey,
createIdempotencyFingerprint,
} from '@axiomnode-lab/guard/idempotency';
const store = new MemoryIdempotencyStore();
const fingerprint = createIdempotencyFingerprint({
method: request.method,
target: request.url,
contentType: request.headers['content-type'],
body: rawBody,
});
const status = await claimIdempotencyKey(
request.headers['idempotency-key'],
fingerprint,
{ store, ttlMs: 86_400_000, scope: user.id },
);
switch (status) {
case 'missing-key':
case 'invalid-key':
// Respond 400: the header is required and must be visible ASCII up to 255 bytes.
break;
case 'accepted':
// Process the operation.
break;
case 'replay':
// Same key and same request semantics.
break;
case 'conflict':
// Same key reused for a different request.
break;
case 'capacity':
// Store is full; fail closed rather than silently evicting a live claim.
break;
}Always pass scope (the authenticated user or tenant) so two callers presenting the same key never collide. The idempotency module stores claim state, not your application response or database transaction result. If you need full response replay, persist that result in application-specific durable storage.
import {
MemoryRateLimitStore,
checkRateLimit,
createRateLimitHeaders,
getClientIp,
} from '@axiomnode-lab/guard/rate-limit';
const store = new MemoryRateLimitStore();
// Only trust X-Forwarded-For hops added by your own proxies.
const clientIp = getClientIp(request.socket.remoteAddress, request.headers['x-forwarded-for'], { trustedProxyCount: 1 });
const result = await checkRateLimit(`ip:${clientIp ?? 'unknown'}`, {
limit: 60,
windowMs: 60_000,
store,
});
const headers = createRateLimitHeaders(result, {
policyName: 'api',
});
if (!result.allowed) {
// Return HTTP 429 with the generated rate-limit headers.
}Use a Redis-backed store for distributed or high-volume services.
import { requireEnv } from '@axiomnode-lab/guard/env';
import { redactSecrets } from '@axiomnode-lab/guard/logging';
const env = requireEnv({
PORT: { type: 'port', default: 3000 },
API_URL: 'url',
CORS_ORIGINS: 'list',
REQUEST_TIMEOUT: { type: 'duration', default: 10_000 },
MODE: {
type: 'string',
allowed: ['development', 'staging', 'production'],
},
});
// env.PORT is a number, env.CORS_ORIGINS is a string[], env.REQUEST_TIMEOUT is milliseconds.
const safeEvent = redactSecrets({ err, req }, {
paths: ['req.headers.x-api-key', 'users.*.profile'],
});requireEnv() returns a frozen object whose types follow the schema. redactSecrets() returns a redacted copy without mutating the original; it keeps Error name/message/stack/cause, redacts camelCase keys such as accessToken, and recognises JWTs, provider API keys and URL-embedded passwords inside strings.
Installing the package exposes the axiomguard command.
npx axiomguard scan .Useful output modes:
npx axiomguard scan . --json
npx axiomguard scan . --sarif --output axiomguard.sarif
npx axiomguard scan . --exclude 'fixtures/**' --max-file-bytes 500000
npx axiomguard scan . --write-baseline .axiomguard-baseline.json
npx axiomguard scan . --github-annotations --no-fail
npx axiomguard rulesThe scanner reports the finding type, file, line and a non-secret fingerprint. It intentionally does not print detected secret values. Exit codes: 0 clean, 1 new findings, 2 error.
Run it without installing Node:
docker run --rm -v "$PWD:/workspace:ro" ghcr.io/axiomnode-lab/axiomguard scan /workspaceSee Scanner and GitHub Action.
The root package re-exports the complete public API:
import {
createApiKey,
safeFetch,
redactSecrets,
} from '@axiomnode-lab/guard';Focused subpath imports are also supported:
import { createApiKey } from '@axiomnode-lab/guard/api-keys';
import { safeFetch } from '@axiomnode-lab/guard/fetch';
import { redactSecrets } from '@axiomnode-lab/guard/logging';Subpath imports make the capability being used explicit and are useful when an application only needs one part of the toolkit.
| Import | Purpose |
|---|---|
@axiomnode-lab/guard |
Complete public API |
@axiomnode-lab/guard/api-keys |
API key generation, hashing, verification and masking |
@axiomnode-lab/guard/webhooks |
GitHub, Stripe, Slack, Meta and Standard Webhooks verification with replay protection |
@axiomnode-lab/guard/crypto |
Secure tokens, constant-time comparison, generic HMAC signing and verification |
@axiomnode-lab/guard/request-policy |
Fetch Metadata and Origin request filtering |
@axiomnode-lab/guard/idempotency |
Idempotency keys, fingerprints and claim stores |
@axiomnode-lab/guard/fetch |
Redirect-aware guarded outbound fetches |
@axiomnode-lab/guard/web |
URL, DNS, IP and redirect safety helpers |
@axiomnode-lab/guard/rate-limit |
Rate limiting and response-header helpers |
@axiomnode-lab/guard/cookies |
Secure cookie serialization |
@axiomnode-lab/guard/cors |
CORS policy helpers |
@axiomnode-lab/guard/csrf |
Signed CSRF tokens |
@axiomnode-lab/guard/headers |
Defensive headers and CSP helpers |
@axiomnode-lab/guard/presets |
Security-header presets |
@axiomnode-lab/guard/env |
Environment validation |
@axiomnode-lab/guard/logging |
Secret redaction and PII masking |
@axiomnode-lab/guard/filesystem |
Safe paths and filenames |
@axiomnode-lab/guard/scanner |
Secret scanner API and SARIF conversion |
@axiomnode-lab/guard/adapters/express |
Express middleware |
@axiomnode-lab/guard/adapters/fetch |
Web-standard Request/Response handler for edge and serverless runtimes |
@axiomnode-lab/guard/adapters/fastify |
Fastify hook |
@axiomnode-lab/guard/adapters/hono |
Hono middleware |
@axiomnode-lab/guard/adapters/redis |
Redis-backed replay, rate-limit and idempotency stores |
AxiomGuard provides security primitives; it is not a replacement for the rest of an application's security architecture. In particular, it does not replace authentication or authorization, a secrets manager, password hashing with Argon2/scrypt/bcrypt, outbound firewalling, durable transaction storage, or application-specific threat modeling.
Read SECURITY.md for vulnerability reporting and THREAT_MODEL.md for design boundaries.
- Node.js: 20, 22 and 24 on Linux, macOS and Windows. Node 20 has reached end-of-life upstream and will be dropped in a future minor.
- TypeScript: 5.x; declarations are emitted with
NodeNextresolution. - Versioning: semantic versioning, pre-1.0. A minor release may tighten a security default; every such change ships with a migration note in UPGRADING.md.
- API reference
- Upgrading between versions
- Comparison with other libraries
- Examples
- API protection
- Framework adapters
- Safe fetch
- Scanner
- GitHub Action
- Security policy
- Threat model
- Changelog
Contributions are welcome. See CONTRIBUTING.md before opening a pull request.
For vulnerability reports, follow SECURITY.md rather than opening a public issue.
MIT — see LICENSE.