Native ClickUp integration with two-way sync
ClickUp now runs through Specship’s native OAuth integration. Tasks can trigger the agent from ready statuses or pickup tags, sync progress comments and PR links back to the task, re-trigger from replies, and move to Done automatically after the merged PR lands.
Continuous, dependency-aware backlog execution
Specship can now work across a related backlog instead of treating every ticket as isolated. Teams can create tickets on the fly while the system keeps executing the queue; dependency- ordered tickets wait for prerequisite pull requests, then unblock automatically as merges land.
Multi-engine coding runs
Private-beta workspaces can run supported engines such as Claude Sonnet 4.6 and OpenAI Codex GPT-5.5 where enabled, so teams can choose the engine that fits the ticket and review policy.
Real test suites, isolated runners, and A-F grading
Jobs run the repo’s actual test tooling, including Vitest, Jest, Pytest, Go test, and per-run Postgres for migration-backed work where configured. Before merge policy runs, PRs receive an A-F quality grade against the ticket’s acceptance criteria.
Rebase before PR and Context Lake retrieval
Runs now refresh against main before opening pull requests and retrieve relevant repository context from Context Lake so tickets start with the files, conventions, and decisions they need.
Founder Mode (private beta)
New per-repo policy that classifies every PR by risk (green / amber / red) and auto-merges the green tier on passing checks. Plain-English changelog replaces the diff-review queue for non- technical founders. Invite-only.
Daily budget cap + hard pause switch
Set a per-project daily token spend cap. When you hit it, every in-flight job pauses, a Slack / email ping fires, and the dashboard surfaces the breakdown. No surprise bills, no "why was my AI bill $400 this morning" story.
Mission Control dashboard
Terminal-aesthetic single page that surfaces every active worktree, the live log console, per-ticket spend, queue depth, and integration health. Designed for the founder who has ten tabs open and one minute to look.
Per-comment PR replies on Gemini reviews
When Gemini Code Assist or CodeRabbit drops review comments, the agent now reads them individually and posts a structured fixed / won’t-fix reply under each comment. Replaces the previous single "here’s the response" mega-comment.
Iteration loop on PR comments
Comment on an open PR — the agent picks it up, iterates on the existing branch (not a new one), re-runs tests, and pushes a follow-up commit. Active feedback log streams to the dashboard while it works.
Searchable GitHub repo + ClickUp workspace pickers in onboarding
Replaced the "paste your repo URL" step with live, filterable lists of repos / workspaces the user can actually access. Much faster for anyone with more than a handful of repos.
Managed repo creation under user’s account
For solo founders who don’t have a repo yet, Specship can create a private GitHub repo under their own account during onboarding with scoped access. They own it, we never do — but the agent gets a writable home base from minute one.
Integration retry flow when OAuth expires mid-build
If an integration token expires while a job is running, the worker now pauses the integration call, preserves the branch state, and re-queues an auth refresh instead of hard-failing the whole ticket.
OSS triage queue
New flow for open-source maintainers: every incoming GitHub issue is scored by Haiku for fixability, then dropped in a triage queue you explicitly approve from. The agent never acts on an un-approved issue and opens PRs as drafts only.
Real-time log streaming
Live SSE-backed log console on the dashboard. Every TDD step, every commit, every test invocation streams in real time — capped height with expand-on-click for long traces.
Visual specs from screenshots
Drop any image (wireframe, screenshot, sketch) into a ticket comment — the agent reads the image, maps it to your existing component library, and matches the layout. Asks if it’s unsure instead of guessing.
First public release
Specship 0.1 — spec-driven TDD agent, GitHub + ClickUp workflow, dashboard, ticket webhook ingress, and the first batch of private-beta workspaces.