deny-sh-integrations-core

The framework-agnostic security core behind every deny.sh agent-framework adapter. You normally don't use this directly: install deny-sh-langchain or deny-sh-vercel-ai instead.

It exposes one primitive, createVaultResolver, which:

1. Resolves a credential from the deny.sh managed vault and decrypts it server-side, inside a closure.
2. Runs your use(secret, args) callback (the only place the plaintext exists).
3. Fail-closed leak sweeps the returned DTO: if the raw secret appears anywhere in it (string, nested object/array, Map/Set, object key), the resolver throws and returns nothing. The secret never crosses the boundary.

import { createVaultResolver } from 'deny-sh-integrations-core';

const resolve = createVaultResolver({
  label: 'stripe-prod',
  password: process.env.VAULT_PW!,
  use: async (secret, args) => ({ /* narrowed DTO */ }),
});

const dto = await resolve({ id: 'in_123' }); // throws DenyLeakError if dto contains `secret`

API

• createVaultResolver(config) → (args) => Promise<DTO>
• isNarrowed(value, secret) → boolean (the sweep predicate)
• DenyToolError — vault fetch / use() failure, secret-scrubbed
• DenyLeakError — fail-closed leak sweep tripped

Config: { label | id, password, use, clientOptions?, leakSweep?, client? }. Inject client (a { vaultGet, vaultGetById }) to mock the network in tests.

Apache-2.0. Part of deny-sh-integrations.