BreakpointHeroku introduces Team Authorizations, allowing admins to create team-owned API tokens and OAuth authorizations instead of tying credentials to individual accounts. The feature improves integration continuity, centralized visibility, and offboarding security with one-year token lifetimes and one-time secret display.
Heroku engineering blog
Andy Smith highlights six lesser-known Heroku CLI commands for exporting configuration, inspecting PostgreSQL queries, debugging dynos, running detached tasks, enabling autocomplete, and opening add-on dashboards. The examples show how these commands can reduce context switching and streamline common operational workflows.
Heroku has made Fine-Grained Access Control available to all customers, replacing fixed roles with capabilities that can be scoped to organizations, teams, pipelines, and apps. The post explains least-privilege permissions, auditability, CLI commands, Platform API integration, and migration requirements.
Moving fast should not mean breaking things for your users. The post explains why a staging environment matters and how Heroku pipelines can provide a near-production replica while keeping the deployment workflow clean and automated.
Beginning in version 11.8.0, Heroku CLI authentication credentials will be stored in the system keychain by default for improved security. The post explains how the credential manager uses OS-native secure storage while preserving existing developer workflows.
Heroku Connect is a fully managed, bidirectional data sync service between Salesforce and Heroku Postgres that lets developers read and write Salesforce data using standard SQL. The post explains how to quickly check the health of the pipeline when fields do not update as expected or data appears to lag.
Heroku CLI v11 completes the migration to ECMAScript Modules and oclif v4, improving startup performance, maintainability, and plugin interoperability. The release also adds Node.js 22 support, semantic themes, accessibility controls, new command namespaces, and workflow improvements.
Heroku outlines its sustaining engineering model and three platform updates: a larger 1 GB compressed slug limit, longer build timeouts, and Heroku CLI v11’s migration to ESM and oclif v4. The changes target more capable builds, faster execution, and improved production reliability.
Heroku raises the maximum compressed app slug size from 500MB to 1GB and increases build compile timeouts for complex applications with large dependency sets. The update provides more deployment headroom while warning that larger slugs can increase dyno boot and scaling times.
This post explains how maintaining cloud native buildpacks (as pack CLI maintainers) ensures infrastructure reliability across platforms and quickly delivers security patches. It highlights how fixes shipped for one platform protect users across Heroku and many other cloud and platform providers. The article gives an inside look at the maintenance work and its cross-platform impact.
The article discusses the challenges of observability as applications scale, where logs and metrics grow across dynos, databases, and third-party APIs. It explains the need to extract signals from noise and understand system health in context to optimize performance. The post describes how full-stack visibility helps monitor service interactions and resource limits effectively.
Heroku explains the CA/Browser Forum’s phased reduction of publicly trusted TLS certificate lifetimes, beginning with a 200-day maximum in March 2026. The post outlines renewal requirements for manually managed certificates and shows how Heroku Automated Certificate Management can automate compliance.
Heroku introduces a flexible standard plan for Managed Inference and Agents, allowing one add-on and API key to access its full model catalog. The update adds Claude 4.6, new open-weight models, Cohere Embed V4, and migration guidance for deprecated Claude versions.
Heroku explains how ephemeral one-off dynos provide isolated code-execution sandboxes for AI agents. The post demonstrates programmatic tool calling through the Managed Inference and Agents API and deployment of open-source MCP servers for Python, Ruby, Node, and Go.
Describes a workflow to reduce context-switching during production incident debugging by consolidating IDE, logs, and error data. Demonstrates how Heroku Managed Inference and Agents (using model context) can speed root-cause analysis and incident resolution.
Heroku’s GitHub Enterprise Server integration replaces personal OAuth credentials with a GitHub App identity, enabling decoupled authentication, granular repository permissions, and reduced service-account overhead. The post explains how the integration supports governed CI/CD pipelines, review apps, Terraform provisioning, and phased enterprise rollout.
Heroku announces a transition to a sustaining engineering model centered on stability, security, reliability, and customer support. Existing customers can continue using the platform without changes, while new Enterprise Account contracts will no longer be offered.
If you’ve built a RAG (Retrieval Augmented Generation) system, you may find many semantically similar results that don’t actually answer the query. This post presents a reference architecture and a Heroku example app that show what comes after vector search to build production-grade AI Search.
Heroku introduces managed reranking models for its Inference and Agents platform, supporting Cohere Rerank 3.5 and Amazon Rerank 1.0. The post demonstrates integrating the /v1/rerank API into a RAG pipeline, using relevance scores and top-N filtering to improve retrieval quality while reducing context and inference costs.
Heroku expands its Managed Inference and Agents platform with Claude 4.5, Nova 2, and multiple open-weight models. The post also introduces a preview of Anthropic’s Messages API, explains SDK authentication and compatibility constraints, and details automatic prompt caching and upcoming model deprecations.