Recurring tasks with custom schedules
Open source · Self-hosted · Apache 2.0
Glide fromidea toshipped.
A new way to build software. AI agents scope the request, write the spec and forecast the date. Your team reviews and decides. No sprints, no story points, no standups.
npx glideship initA Glideship board. One feature, "Search across projects and comments", moves from BUD to production while each agent finishes its step.
- Triage Agent Scoped from Slack
- BUD Agent Spec written
- Design Agent Wireframes ready
- Tech Plan Agent Plan ready
- Smart Assignment Assigned to Lucas: best skills, free this week
- Status Agent Approved and merged to main
- Test Plan Agent 14 tests passing
- Bug Linker Agent Clear to release
- Status Agent Live in production
- Learning Agent Retrospective saved
Less process.
More shipped.
Code is the truth.
Glideship learns from your code, not your wiki. Every time a pull request merges into your main branch, it re-learns what changed and shares it with every agent. So specs, plans and tests match what the code really does, not what someone remembers.
See how knowledge syncs
XP for shipping. Not for activity.
Whatever you measure, people learn to game. So Glideship rewards just one thing: work that reaches production. The game and the goal are the same.
- Outcomes, not activityXP lands when work is merged, reaches UAT and goes live in production. Never for ticket counts, hours logged or lines of code.
- Shared on shippingWhen a BUD ships, the XP is shared by everyone who contributed commits or pull requests. Helping a teammate never costs you.
- Recognition, not surveillanceLeaderboards celebrate what shipped. Streaks and mini-games keep things fun, but they never count toward work XP.
- Levels unlock rewardsGood outcomes earn skill points to spend on unlocks and badges. Fun stays fun, and reviews stay reviews.

Bring your backlog.
Bring your AI.

Bring your Jira. Then leave it behind.
Connect Jira Cloud and turn your existing issues into BUDs in minutes. From there, the agents run the process.

Bring your own AI.
Run the agents on Claude Code, GitHub Copilot, OpenAI Codex or a local Ollama model. With Ollama, nothing leaves your network. Your coding assistant also connects through MCP, the open standard for plugging AI tools into apps, to read your team's context and draft BUDs. Every change is saved, so you can undo it in one click.
Ship the work. Skip the ceremony.
One command. Needs Docker and Node.js 18 or later. Your code and data stay on your machine. npx glideship init