deployments

Deployments

This monorepo has three deployment paths:

• telecine -- GCP Cloud Run services deployed via Pulumi; deploy is gated on CI (integration, playwright, typecheck) passing via a PR merged to main
• elements -- npm packages published by pushing a git tag
• skills -- public documentation published to the skills remote via scripts/push-skills

Scripts

telecine and elements are sibling repos cloned alongside the monorepo under ~/Editframe/worktrees/<branch>/. Push to them with standard git push origin from within the sibling repo directory. There are no subtree push scripts.

Script	Purpose
scripts/push-skills	Generate + push skills to remote
scripts/wait-for-telecine-action	Poll telecine deploy CI (30s interval)
scripts/wait-for-elements-action	Poll elements release CI (30s interval)
scripts/wait-for-github-action <repo>	Generic CI poller for any repo
scripts/gh-logs <repo> [run-id]	Get failed job logs for a CI run
scripts/deploy-info telecine	Services, resources, routes, secrets
scripts/deploy-info elements	Packages, release pipeline

Quick Reference

Action	Command
Deploy telecine	Push branch → open PR → CI must pass → merge to main → scripts/wait-for-telecine-action
Push telecine branch	git push origin <branch> (from worktrees/<branch>/telecine/)
Deploy specific services manually	telecine/scripts/build-and-push web scheduler-go
Deploy all services manually	telecine/scripts/build-and-push-all
Run Pulumi directly	cd telecine/deploy && pulumi up
Prepare elements release	elements/scripts/prepare-release <version> (from sibling elements repo)
Bump elements to beta	elements/scripts/prerelease
Publish elements manually	elements/scripts/publish
Publish skills docs	scripts/push-skills
Tag skills release	git push skills $(git rev-parse skills/main):refs/tags/v<version>
Check CI status	scripts/gh-logs editframe/telecine
Poll CI until done	scripts/wait-for-telecine-action

Telecine (GCP Cloud Run)

Telecine runs on GCP Cloud Run, managed by Pulumi (TypeScript) from telecine/deploy/. Docker images are tagged with the git SHA and pushed to GCP Artifact Registry. A push to main triggers parallel Docker builds for all services, followed by a Pulumi deployment.

Supporting infrastructure includes Cloud SQL Postgres, Valkey (Redis) on Compute Engine, GCS storage buckets, an HTTPS load balancer with URL-map routing, and Cloudflare DNS.

Skills source files are embedded in the telecine subtree at telecine/skills/skills/ for the web Docker build (Tailwind class scanning + runtime docs serving).

Run scripts/deploy-info telecine for the current list of services, their resource allocations, load balancer routes, and secrets.

See references/telecine.md for deployment architecture and procedures.

Elements (npm)

Elements publishes @editframe packages to npm via Trusted Publishing (OIDC). Pushing any git tag triggers the release workflow. Tags containing "beta" publish with the beta npm dist-tag; all others use latest.

Run scripts/deploy-info elements for the current package list and release pipeline steps.

See references/elements.md for the release workflow and versioning.

Skills (Public Docs)

Skills are public-facing documentation generated from source files in skills/skills/ and published to the editframe/skills GitHub repo. The generation strips human-only metadata (nav, track, sections, api) and keeps only LLM-essential frontmatter.

scripts/push-skills

# Tag with the same version as elements
git fetch skills main
git push skills $(git rev-parse skills/main):refs/tags/v<version>

There is no CI -- the push is the deployment. Tag skills with the same version used for elements releases.

See references/skills.md for the build process and source file conventions.

Troubleshooting & Operations

When investigating production errors, start with the error triage workflow in references/troubleshooting.md: query Error Reporting API for grouped errors first, then drill into Cloud Run logs, then correlate with revisions to determine whether a deploy introduced the issue.

Rollback procedures, CI failure debugging, scaling resources, and secret management are also covered in references/troubleshooting.md.

When to Use This Skill

Use this skill when:

• Investigating production errors or service health issues
• Deploying services or packages
• Investigating a failed deployment
• Rolling back a bad release
• Adjusting service resource limits (CPU, memory, instance counts)
• Adding or rotating secrets
• Understanding how services are routed (load balancer, DNS)
• Preparing or publishing an elements release