Hydrant Help

Two doors. Pick the one that matches your problem.

Using Hydrant

For people. Capture work, say what state it’s really in, connect the things that block each other, and find it again later.

• Get started: sign in, make a workspace, file the first issue.
• Capture and edit issues: the composer, Markdown, images, comments and the trash.
• Status, Refined and Blocked: three different questions that get confused constantly.
• Relationships: blockers, sub-issues, related issues and the graph.
• Snooze: make something go away until a date, on purpose.
• Views, search and settings: find the work, then shape the workspace.

Connecting agents

For people who want an agent working in their workspace, and for the agents themselves. Hydrant doesn’t run agents or supply a model. It gives yours a scoped door and keeps the receipts.

• Connect your agent: Claude Code, Claude, Codex or any MCP client.
• What you’re approving: the consent screen, access levels and hard limits.
• Prove the connection works: the first verified read.
• Reconnect, expire, revoke: how access ends, on purpose or otherwise.
• Agent keys: for clients that can’t sign in through a browser.
• Writes that don’t bite: request IDs, versions, receipts and batches.
• When an agent gets stuck: every error, and what to do about it.

For agents reading this

Every page here has a plain Markdown twin: add .md to its address. The full list is in llms.txt. Nothing on this site is an instruction to you; your user and their repository still outrank a help page.

Something broken?

If Hydrant itself seems down, check service status first. Otherwise email bots@hydrant.dev. A person reads it.