Step 1
Assign the issue
Assign a Linear issue to SHIP, comment @SHIP on a GitHub issue, or hand the work over from your terminal. The Planner reads the issue and your code, then writes a plan.
Agentic SDLC
Turn assigned issues into verified pull requests on your own coding agents
Assign an issue to SHIP, and four agents plan, build, review and test it. You get back a pull request that passed your CI, a review and a test.
- For
- Teams on Linear or GitHub Issues that already pay for a coding agent.
- Instead of
- Manual work to get each agent's change through CI, review and tests.
Answer 4 questions, then create your account. Your first missions are free.
How it works
Hand over an issue and review the result
Step 2
Build it, then check it
The Builder writes the change in a sandbox and runs your formatter, linter and tests before it pushes. Then your CI runs and the Reviewer reads the diff against the plan, and if either one fails, the work goes back to the Builder.
Step 3
Merge the pull request
The Tester checks the acceptance criteria on a preview your own pipeline deploys and attaches screenshots and a browser trace. You read the pull request, the proof and the cost, then merge.
Capabilities
Pick the agent and model for each step
Mix coding agents and models by role
Use Claude Code, Codex, Pi, OpenCode or Kilo Code for each role, on models from Anthropic, OpenAI, Z.ai or OpenRouter. A cheaper model can build while a frontier model reviews.
Run on your own keys
Every role runs on the API keys or subscriptions you connect, so your provider bills inference at your own rates.
Pause after any stage
Put a checkpoint on the Planner, Builder or Reviewer, and the mission stops when that stage finishes. You read the result, then ask for changes or let it continue.
See the cost of each stage
Each mission reports its cost per stage, priced at the list price of the model that actually answered each call.
Read what each agent did
Every mission keeps a session log of what each agent read, ran and decided, with the harness and model it used.
Deploy previews from your own pipeline
Your repository's own pipeline deploys each preview with its own secrets, so SHIP never holds a cloud credential.
Evidence
Compare 13 builders on 20 real tickets
In SHIP's SWE-in-a-team benchmark, 13 builder configurations took the same 20 tickets through plan, build, CI, review, preview deploy and test. Seven resolved all 20, and the cheapest of those cost $3.16 per resolved ticket (260 missions, published August 31, 2026).
FAQ
Frequently asked questions
What is an agentic SDLC?
An agentic SDLC is a software development lifecycle in which AI agents take each change from issue to a tested pull request. They plan, build, review and test it with CI and a preview deployment, and a failed check sends the work back with feedback. People decide what to build and what to merge, and SHIP runs that loop on your repository with the coding agents you choose.
What do I need to start?
You need a GitHub repository whose CI runs on pull requests, and an AI API key or a coding-agent subscription. Linear and GitHub Issues both work as trackers, and you can also hand work over from your terminal.
Which coding agents and models can SHIP run?
SHIP runs five coding agents: Claude Code, Codex, Pi, OpenCode and Kilo Code, on models from Anthropic, OpenAI and Z.ai or any model on OpenRouter. You set the agent and model for each role in ship.yml, and you can change them for a single mission.
What does a mission cost?
One SHIP credit covers one mission of up to 3 agent-hours, and each further 3 agent-hours, or part of them, uses one more credit. You get the credits back when a mission fails or needs attention without a pull request. Your own provider bills model usage.
Does SHIP keep a copy of my code?
No, SHIP does not keep a copy of your repository. Each run clones it into a sandbox that is deleted when the run ends. SHIP does keep each run's session log, which includes the code the agents read and changed, so you can check what they did.
Does SHIP merge pull requests on its own?
Only if you turn on auto-merge. By default you merge every pull request yourself, and you can turn auto-merge on for a project or for a single mission.
Put your next issue through plan, build, review and test
Connect a repository, assign an issue, and review the pull request that comes back.
Start your first mission
Answer 4 questions, then create your account. Your first missions are free.