Docs
Start where you already work.
Harbor keeps what your team decides and serves it back to every agent your team runs. These pages are the whole loop: wire it up, feed it, and read what it actually did.
What Harbor is
Your team decides things all day: in pull request reviews, in Slack, in a doc nobody re-reads. Harbor reads the tools you already use, turns what was settled into short rules a person approves, and hands those rules to your agents at the moment they start working. Then it counts which ones the agents were given and which ones they actually used, so a rule nobody cites can be retired instead of accumulating.
Nothing an agent proposes reaches the team without a person approving it. That is the same gate on every surface below.
Which of these is yours
I write code in a terminal.
Start with harbor context. It needs no account, no token and no network, and it prices what your agents already carry on every turn.
CLI →I use Claude or ChatGPT and install nothing.
Connect Harbor as an MCP server and your assistant can search and cite the team graph in the app you already have open.
Agents and MCP →I am setting this up for a team.
Six steps from an empty workspace to a rule an agent was served and a number saying so.
Quickstart →Everything here
Four moves, and they are the loop in order. Something goes in, an agent uses it, you decide what stays, and you can see whether any of it worked.
Orient
what this is, and which path is yours
- QuickstartTen minutes from nothing to your first served rule.
Wire
get your agents, and your team, reading and writing
- CLIharborloop in full: every command, every flag. The first one needs no account.
- Agents and MCPClaude Code, Codex, Cursor and any MCP client, and what each one can honestly do.
- Ask from SlackOne command, a cited answer only you can see, and the three permissions Harbor does not hold.
Feed
decide what goes in, and who it is for
- SourcesNine connectors, and what each one reads. The column that matters is what it refuses.
- Knowledge and reviewFive kinds, one setting that decides what waits for a person, and what leaves the brain.
- Spaces and accessFive roles, thirteen capabilities, and which of them reach into a Space uninvited.
Prove
find out whether any of it worked, and what your agents may not do
- Served vs citedHow many of your rules reached an agent, and how many it actually used.
- DistributionShip a document to every repo that needs it, and seven states saying whether it landed.
- GuardrailsThe commands your agents may never run, what a refusal hands back, and the three switches that all have to be on.
The page on what we store, where it lands and what we never sell is being written now. If you need it today, ask and we will send you what we have.