A tracker your agent
can ship from.
Hydrant wires issues, decisions, notes, and dependencies into one graph — so an agent can pull context, ship work, and leave a trail you can read.
An agent ships from a briefing, not a ticket.
A Brief is the issue, every decision it touches, every note it cites, and the dependency footer — assembled into one markdown payload your agent can read in a single pass.
Regenerate it any time the underlying graph moves. The same brief shows up over MCP, in your shell, and in CI.
feat/pat-handoffRevised 4m ago by Henry. Two issues depend on this decision and are now pending.
When the call moves, every issue that leaned on it knows.
Decisions are versioned. Revise one, and every issue linked to it goes pending until someone acknowledges the new version still holds — or dismisses it as not relevant.
No silent drift. No agent shipping against a decision the team already moved past.
Ready to ship is a checklist, not a vibe.
You decide what makes an issue shippable in your space — description, labels, the right cycle, the right milestone. Every issue carries its own readiness alongside its status.
Triage stops being a judgment call. Your agent stops guessing whether an issue is ready to pick up.
- pivia Claude Codeset dependencies on HYD-88set_dependencies·2m ago
- HKHenryrevised DEC-24update_decision·4m ago
- pivia Claude Codeadded a comment to HYD-71create_comment·7m ago
- MRMarieacknowledged revalidation on HYD-89ack_revalidation·12m ago
- pivia Claude Codecreated HYD-90create_issue·21m ago
Your agent is a named account, not an anonymous token.
Claude Code, Cursor, Codex — each instance gets its own identity in your space. Every mutation carries the actor and the tool that did it.
You read what your agent did the same way you read your own commits — by author, with a diff, on a timeline.
Description prose isn't the source of truth.
Dependencies, sub-issues, and links between issues, decisions, and notes are first-class records. Replace them in one atomic call, not by editing a paragraph an agent hopes is up to date.
Briefs read the graph. Readiness reads the graph. Preflight reads the graph. One source, every surface.
One model. Wherever your agent already works.
The same issues, decisions, and graph — over MCP, in your shell, through REST, or in the web cockpit.