Changelog

Project pausing, deeper metrics

Metrics page for a public service in the Specific dashboard showing request counts, 5xx error rate, and p50/p95 latency charts

  • You can now pause a project from its dashboard settings and resume it later. Pausing stops services and scheduled jobs across all environments while preserving databases, files, and configuration, reducing costs to storage only until you resume.
  • You can now inspect request counts, 5xx error rates, and p50/p95 time to first byte on a public service’s Metrics page. Sort routes by traffic, errors, or latency, then select a route to narrow the charts.
  • We improved database metrics with average compute usage in compute units (CU) and highlighted periods when your database has scaled to zero, making active and idle usage easier to understand.

Agent auth, referrals, better metrics

  • Your coding agent can now authenticate and deploy without you creating an account first, using specific login --agent. Agents can deploy one free-tier project, which you can take ownership of with specific claim within 48 hours. Unclaimed projects are deleted.
  • You can now earn $100 in usage credits when an eligible team you invite subscribes to a paid plan, with up to 10 rewards. Copy your referral link from user settings and track your rewards there.
  • We improved the dashboard’s service metrics with drag-to-zoom, custom time ranges, and deployment markers. CPU and memory charts show peak and average usage across instances, making changes easier to investigate.

More expressive migrations, dashboard improvements

Deployment detail page in the Specific dashboard showing separate Build and Deploy stages with per-service progress

  • Your coding agent can now use more built-in Reshape migration actions to validate data with CHECK constraints, optimize queries with partial and expression indexes, and control what happens to related rows when records are updated or deleted.
  • We redesigned the dashboard’s deployment detail page with separate Build and Deploy stages, clearer per-service progress, and more prominent failures and app links.
  • We streamlined Secrets & Config with inline editing, clearly marked unsaved changes, and a Save & redeploy action. You can review several edits before applying them.

Service alerts, usage breakdowns, preview redeploys

Alert email from Specific: Elevated 5xx responses on the web service, with the error rate and a link to the dashboard

  • You can now monitor service errors, slow responses, failed health checks, and crashes from the dashboard’s Alerts page. Open an alert to see its history and relevant route breakdowns, or copy a debugging prompt for your coding agent. Pro and Scale include email notifications, with control over who receives each alert.
  • On paid plans, you can now see which projects and environments contribute to your usage costs in organization settings. Switch the chart between project and resource breakdowns, and expand a project to inspect its environments, with previews grouped together.
  • Your coding agent can now redeploy an existing preview with specific deploy --environment <name|id>, without creating another preview or changing your checkout’s saved environment.
  • We also improved deployment reliability, reducing interruptions during rollouts and preventing slow image pulls from cutting into your app’s startup health-check time.

Headless API keys, reliable local workflows

  • You can now create API keys in user settings for CI, cloud coding agents, and other environments where browser login is unavailable. Keys can have full or read-only access, an expiration date, and can be revoked from the dashboard.
  • specific exec now follows your dev.env overrides when attached to specific dev. Commands use the same dependencies and values as the running service instead of starting resources you intentionally disabled.
  • Public URLs created by specific dev now reliably forward file uploads and other streamed request bodies without truncation.
  • specific status now distinguishes the tier that is actually running from a tier change that failed to deploy, making unsuccessful scaling changes easier to diagnose.

Database credentials, deployment retries, faster previews

Access credentials tab in the dashboard's Database view with a named credential

  • You can now create named database credentials from the dashboard for local development, CI, and external tools. Each credential is separate from your deployed services and can be rotated or revoked independently.
  • Your coding agent can now retry a failed or cancelled GitHub deployment against the same commit with specific deployment retry. If you omit the deployment ID, it retries the latest deployment for the selected environment.
  • Preview builds now reuse work from their parent environment, making first preview deploys faster when they share dependencies and build steps.
  • Dashboard analytics now provides more stable and accurate traffic counts. Charts consistently align time periods and exclude bots, automated probes, and speculative page fetches.

Faster previews, GitHub deployment policies, agent feedback

  • You can now enable partial previews on an environment in the dashboard, so preview deploys only build services affected by your changes and use the parent environment for everything else.
  • After connecting GitHub, you can turn off Allow deployments from CLI in an environment's settings to block direct specific deploy attempts. Automatic GitHub deployments continue as before.
  • Your coding agent can now send us product feedback with specific feedback "..." after you turn on Allow feedback from the CLI in user settings.
  • Your coding agent can now see dashboard-managed service scaling and rollout policies in specific status and deployment details.
  • Automatic GitHub deploys now honor standard commit-message skip instructions such as [skip ci].

Protected identity, database access, deploy polish

  • Protected environments can now tell your app who signed in. Specific sends an X-Specific-Token on every request, and your coding agent can verify it with endpoint.<name>.token_secret in specific.hcl.
  • Database pages in the dashboard now have an Access tab for connection details.
  • We also shipped a round of improvements around deployment diagnostics and build reliability, so deploys are easier to inspect and more resilient when a build gets interrupted.

Public buckets, preview overrides, more Postgres extensions

Coding agent making a storage bucket public in specific.hcl

  • Your coding agent can now make a storage bucket public with public = true in specific.hcl, and reference the anonymous read origin as storage.<name>.public_url in env vars. Uploads and deletes still require credentials.
  • You can now override secrets and configs for preview environments from the dashboard. Overrides apply on top of the values copied from the parent environment, so previews can use test API keys while production keeps the real ones.
  • You can now start a subset of your project with specific dev --services. It runs only the services you name plus everything they depend on.
  • Specific Postgres now supports the btree_gist, unaccent, and pgcrypto extensions.

Deployment diagnostics, cron lists, workflow controls

Coding agent checking the latest deployment state with the Specific CLI

  • Your coding agent can now inspect deployment failures from the CLI with specific deployment show. It defaults to the latest deployment, shows failed builds and pending actions, and keeps build output hidden unless the agent passes --output.
  • The dashboard Crons view is easier to scan. Instead of jumping straight into one scheduled job, it now shows all jobs with their schedule, cron expression, and latest run in one place.
  • The dashboard now lets you select and terminate multiple workflows at once, so cleanup is faster when several runs need attention.
  • Deployment history is easier to audit. GitHub-triggered deploys now show commit titles in the deployment list, full commit messages on deployment details, and links back to the commit.
  • Your coding agent gets clearer deploy and preview status from specific status. It now explains PR preview requirements, failed latest deploys stay visible, and scheduled jobs, preview labels, and image builds recover more reliably.

Database queries, preview controls, safer deploys

Coding agent querying a Specific Postgres database from the CLI

  • Your coding agent can now query your Specific Postgres databases from the CLI. With specific query --db <name>, it can inspect live data while debugging. Database queries are read-only, so it cannot accidentally change production data.
  • Preview environments can now use their own per-service compute sizes. Set the preview size from the dashboard before opening a preview deploy, so heavier services can get enough CPU and memory without scaling the base environment.
  • Preview volume clones can now run a reset command when they are created, so copied volumes can be prepared for preview testing before services start.
  • We shipped a round of deploy reliability improvements. Deployment history is easier to scan in the dashboard, unchanged volume-backed services avoid unnecessary redeploys, out-of-memory deploys now show a clearer error, and failed preview redeploys can be retried from the dashboard.

Service console and preview environments

Service console in the Specific dashboard

  • You can now open a console for any running service from the dashboard. Pick a running instance and get an interactive shell without leaving Specific, which makes it easier to inspect containers, run one-off commands, and debug deploys.
  • Preview environments now use copy-on-write clones of your service volumes, so disk-backed state like uploaded files is available for testing instead of starting empty.
  • Preview environments are easier to debug from the CLI. specific status now lists preview environments with their URLs and IDs, and specific query --environment <name|id> can run the same log and metric queries against a preview deploy.

Scheduled jobs, one-click DNS, graph databases

Scheduled job runs in the Specific dashboard

  • You can now define scheduled jobs in specific.hcl with cron blocks. Each run shows up in the dashboard with its own logs, CPU/memory metrics, and a Run now button.
  • Custom domains are easier to set up. When your DNS provider supports Domain Connect, the dashboard now opens a provider-hosted setup flow with the required record prefilled, while manual DNS instructions remain available for everyone else.
  • Specific Postgres now supports ltree and pgrouting in local development and production, so apps that need hierarchies, routing, and network queries can use the same extensions everywhere.

Project health in the dashboard

specific health check

  • The dashboard now flags unhealthy projects in the sidebar. Open the health indicator to see recent service issues across your environments and jump straight to the affected service.

Let your agent debug production

specific query demo

  • Your coding agent can now debug your production apps. With the new command specific query, your agent can fire ad-hoc SQL against your live logs, traces, and runtime metrics.

Health checks, faster deploys, volume metrics

health check configuration

  • Your coding agent can now configure health check endpoints in specific.hcl, which we will use to ensure your deployments don't break anything.
  • Service pages now show a volume usage chart, so you can tell at a glance how much storage you're using.
  • Environments can now be renamed from settings.
  • Deployment uploads are now noticeably faster, in particular for large projects.
  • You can now gate PR preview environments behind a GitHub label, in case you don't want one spun up for every PR.

Custom domains for everyone

  • It's now possible to configure custom domains through the CLI, which means your coding agent can handle it for you.
  • We bundled TimescaleDB into our managed Postgres, improving support for scalable timeseries use cases.

Deployment visibility

  • The deployment list now shows the source (git, CLI, or PR) and total duration of each deploy at a glance.
  • Deployment detail pages have live build and deploy timers, so you can watch the seconds tick by.

Storage in the local dashboard

  • When you run specific dev, the local dashboard now has a storage explorer where you can browse the buckets in your local environment.

Command palette & logs

  • Hit ⌘K anywhere in the dashboard to open a command palette and jump to anything with your keyboard.
  • When you run specific dev, the local dashboard now has a log viewer.

Settings, billing, and protected environments

  • We added a "Usage" section to the organization settings which breaks down your usage over time.
  • We added a feature that lets you mark an environment as "protected", which means we will automatically enforce authentication for it without any updates to the code. Perfect for internal apps.
  • You can now configure automatic onboarding for your organization, so new teammates are automatically added without manual invites.

Deployment status

  • Run specific status to get details on the current state of your project in production.

Jobs & workflow controls

  • Use the new Jobs primitive to run one-off commands against any service. Great for one-off scripts and backfills.
  • You can now trigger Temporal workflows manually from the dashboard.
  • You can also terminate a running workflow from the UI, and the timeline view got clearer too.
  • The pg_trgm extension is now available in our bundled dev Postgres.

Ephemeral preview environments

  • We added support for preview environments. Enable it through the dashboard and we will spin up a copy of your infra for each pull request, including a full copy of your databases and storage.
  • You can also run specific deploy --preview from your own computer to get an ephemeral testing environment spun up.

Multi-environment & horizontal scaling

  • Services are now multi-environment-aware, so you can promote a deployment from staging to prod with one click. When you create a new environment, we copy over the secrets and config from a source environment so you don't have to.
  • Environments can now also be deleted
  • Services can now scale horizontally automatically based on load. Open your service in the dashboard and go to "Scale" to configure.

Invites, volume scaling, and a blog

  • You can now resize persistent volumes from the service scale tab.

Deployment metadata

  • Every deployment now tracks its git commit, branch, and source (git push, CLI, or PR), and shows them in the dashboard.

Temporal workflows, billing upgrades, get-started flow

  • We added a Temporal workflow inspector to the dashboard. You can view workflow runs and dive into the details with a split-panel view.
  • The log viewer got severity indicators, JSON parsing, and clickable rows that expand to show full details.

Built for coding agents

  • The specific CLI has been overhauled so that an agent can use it end-to-end. This means you can ask your agent to build something and it will autonomously sign in, spin up a local environment to test, and finally deploy the project to production.

Postgres extensions

  • Our managed Postgres now supports pgvector for vector embeddings.
  • Postgres now includes ICU collation support, so text sorting and search work better across languages.

Teams

  • Organizations now support multiple users. Invite teammates and collaborate on projects by clicking the project switcher in the top-left and then "Organization settings".

Container images & dashboard polish

  • You can now deploy a service from a pre-built container image, not just from source.
  • Builds support a custom Dockerfile if you need fine-grained control.
  • You can view the raw version of your specific.hcl directly from the dashboard.

Persistent volumes

  • You can now attach persistent volumes to your services for stateful workloads, like OpenClaw or databases.

A new tunnel server

  • We rebuilt our tunneling server, so specific dev --tunnel now serves your services snappier and more reliable.

String interpolation in HCL

  • specific.hcl now supports string interpolation, so you can reference other resources inline without annoying workarounds.