SShortSingh.
Back to feed

Why Passwork built a three-tier recovery system instead of an admin master key

0
·44 views

Password manager Passwork has detailed how it designed account recovery without a single 'god mode' credential that could decrypt all stored data at once. The company argues that any backdoor capable of unlocking everything instantly becomes a critical single point of failure, turning one compromised account into a total breach. Instead, Passwork uses a zero-knowledge architecture where the server only stores encrypted data and never holds decryption keys, meaning recovery must be planned before an incident occurs. To handle lost access, the system splits recovery into three isolated tiers, each limited to a narrow task, so no single layer can expose the entire vault. The first tier addresses owner login issues via server-level CLI commands that are disabled by default and require explicit activation, keeping them out of reach of web-only attackers.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Čo je llms.txt a či ho váš web potrebuje

llms.txt je navrhovaný štandard, ide o jednoduchý textový súbor umiestnený v koreni webu, ktorý pomáha jazykovým modelom a AI agentom pochopiť, čím sa web zaoberá. Namiesto navigácie, skriptov a štýlov ponúka čistý prehľad v Markdowne so základným popisom, zoznamom dôležitých stránok a kontaktnými údajmi. Niektoré AI nástroje ho čítajú, no žiadny veľký vyhľadávač zatiaľ nepotvrdil, že ho využíva na hodnotenie stránok. Vytvorenie súboru je lacné a jednoduché, no nenahrádza kvalitný obsah ani SEO. Existenciu súboru možno overiť priamo v prehliadači alebo bezplatným skenom.

0
ProgrammingDEV Community ·

How OpenAI runs PostgreSQL for 800 million users without sharding

OpenAI serves ChatGPT from one PostgreSQL primary that handles all writes, supported by nearly 50 read replicas spread across regions. Rather than sharding its existing Postgres deployment, which would require changing hundreds of application endpoints, the company keeps load off the primary by routing reads to replicas and moving shardable, write-heavy workloads to systems like Azure Cosmos DB. It has also stopped adding new tables to this Postgres deployment and applies strict limits at every layer to prevent a single load spike from causing a wider outage. The article, drawing on OpenAI's January 2026 engineering post, describes how outages typically start with a sudden surge that triggers retries and cascading failures. Practical lessons include adding cache-miss locking, setting lock timeouts on migrations, and auditing write queries with pg_stat_statements.

0
ProgrammingDEV Community ·

Cuenta de pruebas conserva permisos de producción meses después del lanzamiento

Un equipo que probó una integración con una cuenta temporal recibió permisos amplios para consultar pedidos reales. Tras un lanzamiento exitoso, la cuenta siguió activa meses después, sin que nadie recordara quién custodiaba su clave ni qué proceso la usaba. El problema radica en que la fecha de retirada nunca se incluyó en el plan, de modo que desactivarla de golpe podría romper procesos y mantenerla deja un acceso sin responsable claro. El artículo propone registrar para cada acceso de prueba quién lo solicitó, qué permisos tiene, cuándo caduca y cómo verificar su retirada. Además, recomienda validar con permisos mínimos definitivos antes de cerrar el proyecto, ya que un despliegue no está completo mientras sus accesos temporales sigan siendo permanentes.

0
ProgrammingDEV Community ·

Paddle as Merchant of Record Turns Dunning Into a Handoff Problem for Sellers

Sellers using Paddle as merchant of record find that failed-payment recovery depends on managing handoffs rather than controlling every charge attempt. Because Paddle handles the first retries and the statement name, sellers see subscriptions move through past-due, paused, and cancelled states instead of raw payment failures. Duplicate recovery efforts risk charging or emailing customers twice, so sellers are advised to avoid competing with the platform's own retries. Sellers retain responsibilities such as capturing card updates, timing retries, checking live subscription status, and setting customer expectations about the billing name. Reconciling recovered revenue against Paddle's own ledger is also recommended, since the platform serves as the system of record.