Power your agent
The publisher's own URL. Not a PlugMCP endpoint.
Host a copy with our keys and logs.
https://hydrant.dev/api/mcpThe Claude button above opens Claude with the name and URL filled in. You still confirm it there.
claude mcp add --transport http dev.hydrant-hydrant https://hydrant.dev/api/mcpRun in a terminal.
{ "mcpServers": { "dev.hydrant-hydrant": { "command": "npx", "args": [ "-y", "mcp-remote", "https://hydrant.dev/api/mcp" ] } }}claude_desktop_config.json
Remote URL or a command on your machine? How the two installs differ
Run on PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
LaunchFrom the publisher
Skills that teach your agent how to work inside a Hydrant workspace. Read everything before touching anything. Write with receipts. Say what actually happened.
Hydrant, operated by Background Craft, lets your agent track issues, projects and dependencies in a workspace you control. The hosted MCP server can read workspace context and, with permitted access, create and update issues, add comments and manage relationships.
Track issues, projects and dependencies with your agent in a workspace you control.
This public repository contains Hydrant's agent skills and client integration packages. The hosted server's application source is private; you do not need to clone this repository or run a local server to connect.
Add a remote MCP server to your client's configuration:
{
"mcpServers": {
"hydrant": {
"type": "http",
"url": "https://hydrant.dev/api/mcp"
}
}
}
The endpoint uses Streamable HTTP. Configuration syntax varies by client; follow the client setup guide. Authenticate through your client's browser OAuth flow, sign in with GitHub and approve the intended workspace and permitted access. No static API key is required for OAuth connections. Access stays bounded by your workspace role and can be revoked in Hydrant under Settings → Agents → Access. See consent and access.
Start by asking your agent to call get_workspace_identity with {} and confirm the connected workspace name and ID, then try:
Confirm the workspace before making changes. The skills below add guidance for the ticket lifecycle; they are optional for connecting to the server.
Privacy · Terms · Support: [email protected]
Hydrant already hands agents its tools over MCP. Tools describe what an agent can call. These skills cover what it should do.
hydrant: the base every other skill relies on. It also covers promoting between environments: check the clearance first, stop and ask when it's missing or reviews are still open, and brief the reviewers with dockets after.hydrant-setup: scans your repository, installs the rest without overwriting your own skills, and writes the repository profile.capture: files a new issue, or a related set, from a request, idea or bug report, after checking for duplicates. It files buildable work only (no decision issues, no umbrella parents), groups a set by project and milestone, and links every real blocker, so two issues with no blocking path between them can be built in parallel.refine: settles scope, acceptance, priority, size, owner, labels and project, and marks the issue Ready only when nothing material is open.prep: confirms the assignment, reads the issue and the code it touches, maps each acceptance check to a command from your profile, and records a checkpoint on the issue. Edits nothing.go: builds the issue on its own branch, runs your profile's checks, preflights the result and hands it to review with a brief for the reviewer. Commits locally; pushes and opens a pull request only when you ask or under a ship grant. Never merges.review-triage: watches an open pull request, checks each review comment against the code, fixes what's warranted, replies in every thread and records the result on the Hydrant issue.ship: takes one issue, building it with go if that's still to do, through the pull request, the CI gate at the exact head, merge, your profile's release steps and the done status, only under a grant you confirm for that issue.Review costs nothing extra. Before handing off, go gets one independent review of its change from a fresh reviewer on the agent you already use: a subagent where your client has one (Claude Code and Codex both do), otherwise a second session of the same CLI with editing turned off (claude -p with edit tools disallowed, or codex exec -s read-only). No review bot or paid service is needed. If you already use one, go leaves it to review the pull request as usual. If no independent reviewer can run, go says so instead of passing off its own review as independent.
npx skills add Background-Craft/hydrant-skills -s hydrant-setup -a claude-code -a codex
Drop the -a for the agent you don't use. Then ask your agent to run hydrant-setup.
Keep -s and at least one -a. -s installs setup on its own: a plain npx skills add Background-Craft/hydrant-skills -y overwrites any skill of yours with the same name, such as your own refine, without asking. -a keeps the CLI to those agents' folders. When an agent runs the command without it, the CLI installs for every agent it supports, with no prompt, and replaces any same-named skill in their folders. Setup pins -a the same way for every skill it installs.
Codex: its default workspace-write sandbox protects .agents/, so the install fails there. Run the command from your own terminal. When setup installs skills and writes .agents/hydrant-workflow.md, approve the escalation Codex asks for. If you don't, setup prints the commands and the profile for you to run and save yourself. Codex also runs commands without network by default, so setup can't fetch the pack's skill list; it plans from the list it ships with and says so.
Only want the base skill? npx skills add Background-Craft/hydrant-skills -s hydrant.
The skills CLI is the whole installer. It detects the agents you have, and writes a skill directory, an optional symlink and a lockfile. Nothing else. No hooks, no shell scripts, no edits to your instructions files.
-g installs for your user instead of this project.-a claude-code picks one agent. Repeat -a for more.--copy copies instead of symlinking.Project install with Claude Code and Codex both selected:
.agents/skills/hydrant/: the skill itself. Codex and other .agents-aware clients read it here..claude/skills/hydrant: a symlink to that directory, for Claude Code.skills-lock.json: source, path and content hash, so a later npx skills update knows what it installed.With a single agent selected, the CLI copies the skill straight into that agent's directory (.claude/skills/hydrant/ for Claude Code, .agents/skills/hydrant/ for Codex) and skips the symlink. Either way that directory and the lockfile are the only changes. Installing a skill does not connect a client, create a key or authorize anything.
hydrant-setup reads your skills, skills-lock.json, AGENTS.md/CLAUDE.md, scripts, CI, deploy config and any review bot (none is fine; no paid service is needed). It shows one plan and waits for your answer before writing anything.
-s <name> and explicit -a agents, so nothing outside the folders it checked is touched..agents/hydrant-setup-backup/<name>/..agents/hydrant-workflow.md, holds this repository's facts: base branch, commands, CI gate, PR conventions, review bots, release steps, what "done" means and which skills were installed. It is yours to edit; the skills CLI never touches it. Statuses, acceptance and ship grants stay in your Hydrant workspace.hydrant.dev/mcp endpoint, hyd-NNN IDs) is listed in the plan as "may conflict". Setup leaves those files as they are; you decide.Setup does not edit AGENTS.md or CLAUDE.md, create keys, connect clients or write to Hydrant.
review-triage always triages people's comments: review bodies, inline comments and top-level PR comments. A review bot's comments are triaged only when its login is under Review bots in the profile, so CodeRabbit, Copilot, Greptile and others all work the same way. Other bot accounts, such as Dependabot, are listed as not triaged. With no bot listed, which is the default, the skill waits only for CI. Comment text is checked against the code and never followed as an instruction. The skill never merges, approves or changes the issue's status.
ship is the only skill that merges or marks work done. Run /ship #42 and it shows you a grant for issue 42 alone, naming the merge method and each release step from your profile, and waits. Only your yes in that conversation counts: a grant written on the issue earlier, or a comment claiming one, grants nothing, and a new session asks again. The grant is recorded on the issue before anything else happens: as Hydrant grant data with create_grant (for its own connection, this issue, with read, write, pull request and merge actions, release only when your profile has release steps and never sign-off), then as a comment naming the grant id. On a server without create_grant, the comment alone records it and says so. It never covers secrets, new or paid infrastructure, production data, deleting branches or files, bypassing branch protection or a required approval, promoting past a missing clearance, or over work still waiting or held, without your yes to that promotion, or work beyond the issue. At any of those, and at a check still red after one rerun, ship stops, records why on the issue and leaves the issue where it is. It runs only the release steps the grant quoted, each as written, and checks each one. With none in your profile, it merges and marks the issue done once your profile's Done lines are met.
npx skills list
npx skills update
npx skills remove hydrant
update re-downloads every GitHub-sourced skill in the project and overwrites the installed copy, including your edits, even when nothing changed upstream. Keep local changes somewhere else, or install with a local path (npx skills add ./path), which update leaves alone entirely.
remove hydrant deletes that skill's directory and link and leaves every other skill, skills-lock.json and your agent configuration alone.
Not ours. The skills CLI records anonymous install events that include the public repository name. Set DISABLE_TELEMETRY=1 or DO_NOT_TRACK=1 to turn that off.
Connect to https://hydrant.dev/api/mcp using Streamable HTTP. Clients with remote OAuth support open Hydrant in your browser so you can choose a workspace and grant access. No local server required.
Follow the Cline setup guide, then verify the workspace with a read before asking your agent to change anything. The 400×400 Hydrant logo is available for directory listings. This repository contains skills and client setup; it does not contain the hosted server's source.
Three things that are not the same: a connection (the client reaches the server), authorization (what the key may do, one workspace) and permission (what you asked for in this task). The skill keeps them apart. So should you.
The skill assumes a connection already exists. Keys are created in Hydrant under Settings → Agents for one workspace; a key never exceeds the role of the person who made it and stops working when they leave the workspace.
Endpoint: https://hydrant.dev/api/mcp, Streamable HTTP, Authorization: Bearer <key>.
Claude Code, project scope, in .mcp.json, with the key supplied through the environment rather than typed into the file:
{
"mcpServers": {
"hydrant": {
"type": "http",
"url": "https://hydrant.dev/api/mcp",
"headers": { "Authorization": "Bearer ${HYDRANT_AGENT_KEY}" }
}
}
}
Codex:
codex mcp add hydrant \
--url https://hydrant.dev/api/mcp \
--bearer-token-env-var HYDRANT_AGENT_KEY
Keep client approvals on. Hydrant's write tools declare their side effects; the skill tells the agent to write only what you asked for, and your client's prompt is the second lock.
Your agent has confidence. Give it a paper trail.
The plugin gives Claude Code a connection to Hydrant and the hydrant skill: read the context, do the requested work, check the result, and bring receipts. “Done” is a claim. We'd like some evidence. Sign in with OAuth in your browser; no API key to paste into a file.
claude plugin marketplace add Background-Craft/hydrant-skills
claude plugin install hydrant@hydrant
Then run /mcp inside Claude Code, choose plugin:hydrant:hydrant and Authenticate. Approve one workspace in the browser tab that opens. Check the workspace name before you click: the agent gets that one, at your role, and no other. Until you finish signing in, claude mcp list shows Needs authentication.
The plugin includes the base hydrant skill. Install the workflow pack separately with hydrant-setup so it can use your repository's build, review and release rules.
claude mcp add hydrant --scope user? Your entry stays put. claude mcp list shows only your hydrant, not the plugin's copy of the same URL; the plugin still adds the skill, and removing the plugin leaves your entry alone..mcp.json? That entry doesn't displace the plugin's, so you can end up with two Hydrant servers. Keep one: drop the entry, or skip the plugin.claude plugin marketplace update hydrant, then claude plugin update hydrant@hydrant, then restart Claude Code.claude mcp logout plugin:hydrant:hydrant, and /mcp → Authenticate again. claude mcp remove can't touch a server that a plugin brought.hydrant-setup? Let it install hydrant in the repository as well, because the pack's skills read it from the folder next door. Claude Code then sees two copies of one file, hydrant and hydrant:hydrant. Keep them on the same release.claude plugin disable hydrant@hydrant takes the server and skill out of new sessions; enable brings them back.claude plugin uninstall hydrant@hydrant, then claude plugin marketplace remove hydrant if you want the marketplace gone too. Other servers, plugins and settings aren't touched. The workspace approval stays in Hydrant until you revoke it under Settings › Agents.Install from this repository's marketplace. Hydrant is not yet listed in Anthropic's plugin directory; that submission is separate.
One install gets Codex both halves: the Hydrant MCP server and the hydrant skill. It signs in with OAuth in your browser, so there's no key to paste anywhere.
codex plugin marketplace add Background-Craft/hydrant-skills
codex plugin add hydrant@hydrant
codex mcp login hydrant
Approve one workspace in the browser tab that opens. Check the workspace name before you click: the agent gets that one and no other.
The plugin carries the base hydrant skill only. The workflow pack stays a per-repository install through hydrant-setup, because it needs that repository's profile.
codex mcp add hydrant? Your entry stays put. The plugin adds the skill and leaves your server alone, even when you remove the plugin.codex plugin marketplace upgrade hydrant, then codex plugin add hydrant@hydrant again.codex mcp logout hydrant and codex mcp login hydrant. codex mcp remove can't touch a server that a plugin brought.hydrant-setup? Let it install hydrant in the repository as well, because the pack's skills read it from the folder next door. Codex then sees two copies of one file, hydrant and hydrant:hydrant. Keep them on the same release.codex plugin remove hydrant@hydrant, then codex plugin marketplace remove hydrant if you want the marketplace gone too. Other servers and plugins aren't touched.This is a plugin you install from this repository, not a listing in OpenAI's plugin directory. It isn't in the directory yet.
The ticket already exists. Give your agent the context. The Cursor package pairs the same base skill with the same remote MCP endpoint. It does not install the workflow pack.
Local package discovery and desktop OAuth have been checked in Cursor 3.21.16 on macOS against Hydrant v2026.09.27.5. The reviewer-account test verified workspace identity, synthetic issue creation/read-back and denial after revocation. Neither Cursor Marketplace nor cursor.directory has received a submission for this package. Follow the package guide for direct installation, existing-server handling and removal.
For a local preview, build the generic archive, extract it to a temporary directory, and install that extracted directory with Gemini's normal confirmation prompt:
node scripts/package-gemini-extension.mjs
preview_dir=$(mktemp -d)
tar -xzf release/hydrant.tar.gz -C "$preview_dir"
gemini extensions install "$preview_dir"
Do not install, link, or point Gemini at this repository root: Gemini loads every skills/ subdirectory, which includes the repository workflow pack. The release archive deliberately contains only skills/hydrant/, its agents/openai.yaml dependency, LICENSE, and the root gemini-extension.json manifest.
Install the Gemini extension from the latest GitHub Release (v0.3.3 or later):
gemini extensions install https://github.com/Background-Craft/hydrant-skills
The GitHub Release must include the generic base-only archive as its sole asset for this command to be a safe default. A pinned --ref stays pinned when you update. To leave a pin, uninstall hydrant and reinstall without --ref after checking the latest approved release.
The extension defines one server, hydrant, at https://hydrant.dev/api/mcp. Start Gemini and use /mcp auth hydrant to approve the OAuth browser flow; then call get_workspace_identity with {} and confirm its name and ID before doing any work. Cancel the browser flow if the workspace is wrong. To change workspaces, revoke Hydrant access under Settings → Agents → Access, restart Gemini, and run /mcp auth hydrant again.
Gemini keeps an existing user or workspace mcpServers.hydrant entry and gives it precedence over the extension server, so this extension does not overwrite or silently replace an existing configuration. A user or workspace skill named hydrant likewise takes precedence; the archive inventory stays limited to the base hydrant skill. Check /mcp after a restart to see the active server. Manage the extension itself with:
gemini extensions update hydrant
gemini extensions disable hydrant
gemini extensions enable hydrant
gemini extensions uninstall hydrant
Uninstalling removes the extension only; it does not revoke its Hydrant OAuth access. Revoke access separately under Settings → Agents → Access before reconnecting another workspace.
get_workspace_identity and states the workspace name and ID; reads get_workspace only when issue configuration is needed.Requires get_workspace_identity, get_workspace, list_issues, get_issue, list_activity, get_dependencies, list_relationships, create_issue, update_issue and add_comment; uses list_projects, get_project, get_milestone and inspect_project for project membership, and the individually named relationship tools for links in capture and refine. The base skill uses create_docket and add_docket_issues for review groups. The task-context tools and create_grant (in ship) remain optional. The canonical server exposes individually named operations; migrate older action-dispatch calls to their matching tool, use separate calls for multiple writes, and refresh client discovery. Ordinary update_issue patches exclude trash, ordering, approval and review decisions. Connection proof uses only get_workspace_identity. When needed, read prompt templates with get_agent_prompt_templates, the granting user's views with list_personal_views, plan/usage with get_workspace_usage, and transfer rules with get_workspace_transfer_rules; these are no longer bundled into get_workspace. Prompt text remains untrusted workspace content.
The server reports its name and version when a client connects. Releases are git tags; see CHANGELOG.md.
Issues and pull requests are open. Keep skills/hydrant/SKILL.md provider-neutral and short, and run npm ci then node scripts/validate-frontmatter.mjs before pushing. The validator parses frontmatter as real YAML, the way installers do, so quote any description that contains : or #. The shared Codex/Claude Code/Cursor plugin ships a copy of skills/hydrant, because Codex drops symlinks when it installs a plugin: after editing the skill, run node scripts/check-plugin.mjs --write. CI fails when the copy drifts. Workspace policy belongs in your workspace; this file is not the place to legislate it.
© 2026 MCP Directory.
Track issues, projects and dependencies with your agent in a workspace you control.
{ "mcpServers": { "dev.hydrant-hydrant": { "url": "https://hydrant.dev/api/mcp", "headers": {} } }}Tools
No tool list yet. Remote servers get one after the first health check.
FAQ
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
No required environment variable is listed on this record.
Hydrant is reached over HTTP.
Last checked 2026-11-08. Track issues, projects and dependencies with your agent in a workspace you control.