running
HAR-412
Adding the dependency-graph cache so swim lanes stop recomputing per request.
researchplanimplementreview
the loop
One human, steering a fleet of agents
AI made writing code cheap — it didn’t make knowing what to build any faster. Harbour keeps human intent in command of AI execution, one turn of the loop at a time.
-
01
Read the backlog
Point Harbour at Linear, GitHub, or a local store. It reads the whole tree and ranks the frontier — what is actually ready to move.
-
02
Ground a prompt
The right next task becomes a prompt, re-grounded against your code at HEAD — referenced files re-read, stale plans challenged, not trusted blind.
-
03
Dispatch to an agent
Queue it for a coding agent to poll, claim, and run — or fan a whole cohort out at once. One human, a fleet of sessions.
-
04
Verify on evidence
Work lands against real proof — CI, merges, diffs — not the agent’s say-so. Observation shows every run as it happens.
observation
Watch the work happen
Every dispatched run, live: what it is doing, how long it has taken, and the evidence it produced — not a spinner, the actual work.
done
HAR-408
Fixed the 401 refresh loop; token now refreshes once and retries the request.
researchimplementreviewclose-out
queued
HAR-415
Extend the proxy wire contract with a source badge for merged workspaces.
researchplanimplementreview
swim lanes
See the whole board at a glance
Parallel tracks, with the dependencies drawn in. Know what is ready, what is moving, and what is held — before you dispatch.
auth
HAR-390
GitHub App install flow
HAR-408
Token refresh retry
views
HAR-412
Swim lane cache
HAR-419
Roadmap trajectory band
⇠ HAR-412
proxy
HAR-377
Source-neutral wire
HAR-415
Merged-source badge
grounded prompts
Prompts that re-check your code first
14 deterministic templates, chosen by an LLM router and re-grounded against the repo at HEAD before they run.
# Re-ground against current code (staleness check) Before trusting the ticket, list the files it references and git log --since=2026-06-30 — re-read them at HEAD. ## Task Cache the swim dependency graph so lanes stop recomputing per request. Referenced: lib/swim-graph.js, lib/graph-features.js. ## What CI cannot prove The cache invalidates on issue write — verify against a real mutation, not a green build.
try it yourself
See the value for well under $1
Log in and try AI Generated Prompts free — no OpenRouter connection needed to start. Want more? Connect your own OpenRouter key and keep going for pennies.
The default model runs at $0.75 in / $4.50 out per 1M tokens — a single generated prompt costs a cent or two, so dozens fit under $1.
This covers trying AI Generated Prompts — running full autopilot dispatch uses a separate, pricier model and can run long, though even that model stays relatively cheap.
any backend
One cockpit, whatever tracks the work
- Linearthe original backend
- GitHub Issuesfirst foreign backend
- GitHub Projectsboards, read-only V1
- JiraAtlassian sites, token or OAuth
- Localwritable, no tracker
the workstation
Harbour OS
experimentalHarbour is the control plane; Harbour OS is the in-browser workstation it dispatches sessions into. Parent and child, like Apple and macOS — Harbour picks the work, Harbour OS runs it.
Open Harbour OS →$ harbour dispatch HAR-412 --target local → session spawned, agent claimed the item ◐ implement running 4m 12s [evidence] 3 files changed · tests green $ _
the library
Read the work behind Harbour
The papers and essays written while building Harbour — the ideas, the loops, and what it took to get the work done.
- The Harbour Archive (#2). Six months of the project, January to July 2026, laid out as a museum.
- The Folded Loop. An outsider reads the project cold and finds its point: who gets to say the work is done.
- The ladder. Six rungs from asking a chatbot one question to handing agents the backlog, and what it takes to climb each one.
- Does the writing get longer faster than the ideas do? Do agents' reviews grow faster than what they have to say? Yes, by about 1.6×.
- Why does a plan go round plan-review more than once? Plan review sent back 83 of 94 plans, and not because of the design.
- Learning While the Tools Change. Experience, AI skill and the tools change at different speeds. What a team needs most is calibrated distrust.
the archive
Six months, preserved
A museum of the Harbour project — two repositories and the machines that built them, January to July 2026.