Issues

Backend & infrastructure roadmap (pick issues from here, top to bottom)

ParthKapoor-dev/devex#33 · opened

on Sep 13, 2026

Core should talk to Kubernetes with a least-privilege ServiceAccount, not an admin kubeconfig

ParthKapoor-dev/devex#32 · opened

on Sep 13, 2026

Decide how workspace records survive a Redis failure (persistence and backups)

ParthKapoor-dev/devex#31 · opened

on Sep 13, 2026

Workflow hygiene: pin third-party actions, fix the core pipeline filename typo, least-privilege tokens

ParthKapoor-dev/devex#30 · opened

on Sep 13, 2026

Core deploys can't roll back on failure: no health check, :latest image, no uptime alert

ParthKapoor-dev/devex#29 · opened

on Sep 13, 2026

Turn on GitHub's security features and protect main (Dependabot, private reporting, CodeQL, rulesets)

ParthKapoor-dev/devex#27 · opened

on Sep 13, 2026

Production images are built with Go 1.24, which no longer receives security fixes

ParthKapoor-dev/devex#26 · opened

on Sep 13, 2026

Go dependencies with known vulnerabilities that our code reaches (add govulncheck to CI)

ParthKapoor-dev/devex#25 · opened

on Sep 13, 2026

Deleting a running workspace leaves its Kubernetes resources running

ParthKapoor-dev/devex#24 · opened

on Sep 13, 2026

Runner concurrency bugs: SetInactivityPeriod deadlocks, idle timer can fire while connected, data races in PTY

ParthKapoor-dev/devex#23 · opened

on Sep 13, 2026

Markdown twins: a component name inside inline code swallows part of contributing.md

ParthKapoor-dev/devex#22 · opened

on Sep 13, 2026

Backend cleanup: unused code, duplicated constants, emails in logs, stale docs

ParthKapoor-dev/devex#21 · opened

on Sep 13, 2026

Templates pipeline syncs to the production bucket from every branch, and the size check doesn't stop the sync

ParthKapoor-dev/devex#20 · opened

on Sep 13, 2026

Stacks shown or accepted don't match the images that exist (go allowed with no image; landing lists 10)

ParthKapoor-dev/devex#19 · opened

on Sep 13, 2026

Every workspace start pulls :latest images with PullAlways

ParthKapoor-dev/devex#18 · opened

on Sep 13, 2026

Language images copied the runner binary from runner-service:latest instead of the same commit

ParthKapoor-dev/devex#17 · opened

on Sep 13, 2026

runner and mcp didn't compile from a clean checkout (incomplete go.mod, generated protobuf not in git)

ParthKapoor-dev/devex#16 · opened

on Sep 13, 2026

Session cookie stores the GitHub access token, signed but not encrypted

ParthKapoor-dev/devex#15 · opened

on Sep 13, 2026

Core starts with placeholder or permissive defaults when configuration is missing

ParthKapoor-dev/devex#14 · opened

on Sep 13, 2026

Workspace handlers continue after failed Redis writes, and the free-plan limit check is off

ParthKapoor-dev/devex#13 · opened

on Sep 13, 2026

Core Kubernetes calls ignore client errors, have no timeouts, and a half-finished start blocks the next one

ParthKapoor-dev/devex#12 · opened

on Sep 13, 2026

Runner terminal sessions are not cleaned up on disconnect, and the cut/copy clipboard is shared across connections

ParthKapoor-dev/devex#11 · opened

on Sep 13, 2026

Runner WebSocket can drop messages: handlers are registered late and a full buffer discards output

ParthKapoor-dev/devex#10 · opened

on Sep 13, 2026

Workspaces can keep running indefinitely: cleanup depends on one runner callback

ParthKapoor-dev/devex#9 · opened

on Sep 13, 2026

Feature: Add Watcher for File Changes in `/workspaces` and Emit Updates

ParthKapoor-dev/devex#5 · opened

on Jun 22, 2025

Navigation Menu

Repository navigation

Issues

Backend & infrastructure roadmap (pick issues from here, top to bottom)

Core should talk to Kubernetes with a least-privilege ServiceAccount, not an admin kubeconfig

Decide how workspace records survive a Redis failure (persistence and backups)

Workflow hygiene: pin third-party actions, fix the core pipeline filename typo, least-privilege tokens

Core deploys can't roll back on failure: no health check, :latest image, no uptime alert

Turn on GitHub's security features and protect main (Dependabot, private reporting, CodeQL, rulesets)

Production images are built with Go 1.24, which no longer receives security fixes

Go dependencies with known vulnerabilities that our code reaches (add govulncheck to CI)

Deleting a running workspace leaves its Kubernetes resources running

Runner concurrency bugs: SetInactivityPeriod deadlocks, idle timer can fire while connected, data races in PTY

Markdown twins: a component name inside inline code swallows part of contributing.md

Backend cleanup: unused code, duplicated constants, emails in logs, stale docs

Templates pipeline syncs to the production bucket from every branch, and the size check doesn't stop the sync

Stacks shown or accepted don't match the images that exist (go allowed with no image; landing lists 10)

Every workspace start pulls :latest images with PullAlways

Language images copied the runner binary from runner-service:latest instead of the same commit

runner and mcp didn't compile from a clean checkout (incomplete go.mod, generated protobuf not in git)

Session cookie stores the GitHub access token, signed but not encrypted

Core starts with placeholder or permissive defaults when configuration is missing

Workspace handlers continue after failed Redis writes, and the free-plan limit check is off

Core Kubernetes calls ignore client errors, have no timeouts, and a half-finished start blocks the next one

Runner terminal sessions are not cleaned up on disconnect, and the cut/copy clipboard is shared across connections

Runner WebSocket can drop messages: handlers are registered late and a full buffer discards output

Workspaces can keep running indefinitely: cleanup depends on one runner callback

Feature: Add Watcher for File Changes in `/workspaces` and Emit Updates