Free
For a small team trying Trov out.
Free
- 3 seats
- 1 repository
- 2 environments
- 250 MB artifact storage
- 5 agent connections per person
- 300 AI summaries per month
- 1,000 tag checks per month
- 30 tags
trov.dev/acme/#review
Agents start every session cold.
They re-derive conventions and guess at decisions the team already made.
Wikis rot.
Writing back is a chore nobody does after the work ships.
Unchecked agent docs spread mistakes.
One confident error gets read as fact by every agent after it.
Orient, work, record.
01 · Orient
Before touching an existing area, the agent searches Trov and reads the relevant docs and decisions.
02 · Work
The agent builds on what the team decided instead of guessing.
03 · Record
On request, the agent reads what actually shipped from git and GitHub and sends one batch of updates.
Deploy process
Waiting for a verdict since Tuesday.
Reject
Promote
One place for what the team knows.
Open any screen for a closer look.
Docs
A library that stays reviewed
Technical and Product spaces, full version history on every doc, and a heading outline for long pages.
Feed
A timeline of what shipped
Every entry leads with a short brief for people and links to the PRs, commits, and issues behind it. Agents get the full record.
Tickets
A board the whole team files into
Drag cards across Triage, In progress, Testing, and Done, or read the queue as a table by sprint. Comments with @mentions, sub-tickets, and GitHub issues mirrored in as tickets.
Roadmap
Sprints on a calendar
A short narrative with its sprint cards, and a timeline that runs each sprint from start to due date, filled by its done tickets.
My Work
Your day on one page
Your tickets, what's waiting on your review, your recent sessions and handoffs, and the repo at a glance.
Handoffs
Pick up where the last session stopped
An agent leaves a note for the next session or a teammate: what's done, what's next, and the branch. The next session offers it and claims it only when you say so.
Artifacts
Designs and reports, versioned
HTML pages, markdown reports, diagrams, images, and PDFs, linked to the ticket or sprint they came from.
Decisions
Architecture decision records, in their own section of Docs.
Search everything
One box jumps to any ticket, doc, sprint, artifact, prompt, or person. Agents also see pending proposals, labeled.
Prompt Library
Reusable prompts with variables, and a new version on every save.
Repo dashboard
Deploys, CI, drift, and usage from captured data. Unknown reads as unknown, never zero.
Email digests
Daily or weekly, per section, so nobody has to poll the feed.
Light and dark
Or follow your system.
For agents
An MCP server agents connect to directly, and a Claude Code plugin that wires it up with the skills that drive the loop.
Read
queryget_doclist_docsget_feedlist_tagsget_roadmapget_my_workget_eventsget_repo_dashboardlist_people
Contribute
append_feedpropose_doc_updaterecord_session
Tickets
list_ticketsget_ticketcreate_ticketedit_tickettransition_ticketadd_ticket_commentadd_ticket_linkset_ticket_sprintset_ticket_parentassign_ticket
Sprints
list_sprintsget_sprintcreate_sprintset_sprint_activecomplete_sprintadd_sprint_resourcedelete_sprint
Handoffs and prompts
send_handofflist_handoffsget_handoffclaim_handoffexpire_handoffsearch_promptsget_promptsave_prompt
Artifacts and doc images
upload_assetartifact_updateartifact_getartifact_list
Connection
get_connectionswitch_org
Admin
update_plan
Claude Code plugin
10 skills
trovload-contextrecord-sessionmy-workticketshandoffpromptsartifactsread-planupdate-plan
Connect in three steps
- Install the Trov plugin in Claude Code:
/plugin marketplace add AndresL230/trov /plugin install trov@trov
- Run
/mcp, choose trov, then Authenticate. - Your browser opens Trov. Choose how the connection picks an organization, then click Allow — the connection shows up in Settings, under MCP access.
One connection covers all your organizations. When you allow it you choose how it picks one: follow the repository you are working in, or the organizations you tick, one at a time. You can change that later in Settings. Other clients, step by step.
Security, in plain terms
Agents connect by browser sign-in; only hashes of their tokens are stored.
Agents write as their person and can't claim another author.
Agents can only change tickets assigned to their person.
Only a person closes a ticket or completes a sprint. A ticket mirrored from a GitHub issue follows that issue.
Only a person can ratify a decision or an artifact, or publish a prompt.
Artifact pages run in a sandbox, cut off from your session.
GitHub webhooks are signature-verified.
One product, three sizes.
Every plan is all of Trov. They differ in how many people share the organization, and in how much it can connect and store.
Pro
For a team, paid per seat.
$10 per seat / month
Choose Pro
For example, 3 seats × $10 = $30 per month. Plus applicable tax, calculated at checkout. Renews monthly until you cancel; cancel any time from Billing.
- Up to 50 seats
- 5 repositories
- 5 environments
- 5 GB artifact storage
- 10 agent connections per person
- 3,000 AI summaries per month
- 10,000 tag checks per month
- 50 tags
Enterprise
Limits set for the organization by Trov.
Custom pricing
- Unlimited seats
- 10 repositories
- 10 environments
- Unlimited artifact storage
- Unlimited agent connections per person
- Unlimited AI summaries
- Unlimited tag checks
- Unlimited tags
Enterprise limits are where an organization starts: each one can be set for the organization.
In every plan
- An MCP server and a Claude Code plugin for your agents
- Docs and decisions, with a person approving what becomes official
- Tickets, sprints and the roadmap
- The feed, handoffs, the prompt library and artifacts
- The repo dashboard: deploys, CI, drift and usage
- Email digests, daily or weekly
- Integration keys that are write-only and stored encrypted
- Sign-in with GitHub or Google
Questions
Something this doesn't answer? Ask us.
What is Trov?
Trov is shared memory for a software team and its coding agents: the team's docs, decisions, roadmap, ticket queue and a feed of what shipped, in one place. People use it in the browser; agents read and write it over MCP. Agents propose changes, and a person approves what becomes official.
How is it different from a README or a wiki?
A README or a wiki is text someone has to remember to update, and an agent reading it cannot tell a settled page from a stale one. In Trov, agents load what the team already decided before they work and write back what actually shipped when they finish. Every result tells an agent whether it is settled or only proposed, every version is kept, and nothing becomes official until a person approves it. How that works.
Does it work with Claude Code, Cursor and other MCP clients?
Yes. Trov is a remote MCP server at https://trov.dev/mcp, over Streamable HTTP. A client that supports MCP's OAuth sign-in finds everything it needs on its own and opens Trov in your browser: Claude Code, Claude Desktop and claude.ai, VS Code, Codex, and recent versions of Cursor. A client that cannot sign in can send a Trov token you already have as a header. Claude Code also has a plugin with the skills that run the loop. Setup for each client.
What counts as a seat?
A member of the organization, or an invitation that has not been answered yet. A pending invitation holds its seat, so an organization can never accept more people than it has seats.
What does Pro add?
Pro is all of Trov, as Free is: no plan turns a feature on or off. What it adds is room: 50 seats (Free: 3), 5 repositories (Free: 1), 5 environments (Free: 2), 5 GB of artifact storage (Free: 250 MB), 10 agent connections per person (Free: 5), 3,000 AI summaries per month (Free: 300), 10,000 tag checks per month (Free: 1,000) and 50 tags (Free: 30).
How does Pro pricing work?
Pro is bought per seat, at $10 per seat / month. There is no minimum, so a person on their own can buy a single seat. You choose how many at checkout, and add more from Org settings › General when you invite more people, up to 50. The seats you pay for are the seats the organization has. Prices do not include tax, which Stripe calculates at checkout where it applies. Pro renews monthly until you cancel, and an owner can cancel any time from Billing: it stays until the end of the period already paid, with no refunds for partial periods (refund policy).
What happens if we go over a limit?
Nothing is deleted and nobody is removed. Everyone can still read everything. Adding more of that one thing is refused until the organization is back under the limit, and the rest of Trov carries on.
Can we change plans?
Yes. Anyone signed in can create a Free organization (one they own at a time), and its owner upgrades it to Pro from Org settings › General. An owner manages seats, the card and invoices there too. If Pro is cancelled, the organization moves to Free when the paid period ends, and the answer above applies.
Is our data separate from other organizations?
Yes. Everything in Trov belongs to one organization, and in Trov only its members and the agents they connect can read its content. The people who run Trov see an organization's name, members and usage, not its content, and step in only for support, security or legal reasons; the services Trov runs on process it for us (Privacy Policy).
Can we read the source, or run Trov ourselves?
Yes. Trov's source is published on GitHub under AGPL-3.0, and it runs on Cloudflare: one Worker, with D1 for the data and R2 for files, so you can deploy your own copy to your own Cloudflare account. A copy you run is yours to operate: there is no self-hosting guide, and no support for one. The plans on this page are for the hosted service.
How does Enterprise work?
Talk to us about Enterprise. Its seats, repositories, environments, artifact storage, agent connections, AI summaries, tag checks and tags are each set for your organization, and can be changed later.