Convoy orchestrates AI coding agents from a task or spec to a reviewed implementation. Run it from the terminal UI or CLI, choose a model for each step, and keep the reports, costs, and Git history together.

Documentation · Installation · Pipelines · Releases

What Convoy does

• Multi-model pipelines. Implement, audit, test, and review with specialized agents and optional advisors. Choose a workflow →
• Measured quality. Independent scorers and a verified consensus drive bounded rounds of targeted fixes. Quality and goals →
• Worktrees and living specs. Create isolated work, discuss a change, run it against an OpenSpec contract, and review its path to a PR. Worktrees → · OpenSpec →
• Runs you can follow and resume. Watch parallel steps, inspect reports and usage, detach from a live run, and return later. Running Convoy →
• Control over agent actions. Configure permission gates, smart approval, commit checks, hooks, and project instructions. Safety → · Configuration →

Get started

Requires macOS or Linux, Git, and an installed, authenticated OpenCode. The default pipeline uses OpenRouter and OpenAI models. Bun is bundled in the release binary.

curl -fsSL https://github.com/Inakitajes/convoy/releases/latest/download/install.sh | sh

Open convoy in your project for the terminal UI, or start from the CLI:

convoy --prompt-file prd.md       # implement, measure, and fix with full-cycle
convoy -p review                 # report-only review of the current branch
convoy -p ship                   # sync, review, fix, and measure the branch

See installation and updates, provider setup, and CLI usage.

Pick a pipeline

Pipeline	Purpose
full-cycle · default	Implement, audit, test, score, and fix gaps in a bounded loop.
implement	Implement and audit, then produce a run recap.
ship	Sync with the base, review and fix findings, then reach the quality target.
review	Produce findings and a verified score without changing code.
fixer	Reproduce supplied findings, fix them, and verify the result.
hunter	Audit six specialty tracks across five models, then reconcile the findings.

Models, steps, and stopping rules →

Documentation

Guide	Covers
Getting started	Requirements, installation, and upgrades
Pipelines	Built-ins, models, and migration from earlier variants
Quality and goals	Scoring rubric and improve/measure loops
Worktrees	Home, isolation, spin, and close
OpenSpec	Spec-driven runs and the artifact reader
Running Convoy	CLI, TUI, live sessions, history, and resume
Configuration	Agents, advisors, prompts, hooks, and attachments
Permissions and safety	Approval policies, commits, compaction, and publication
Models and providers	Authentication and model gateways
Development	Building, testing, and the source layout