
I Opened Seven Issues for One Feature (On Purpose)
One feature. Seven real issues. Four hours. Zero babysitting.
- A seven-issue dependency graph completed in merge order.
- Detent returned dependent work to Todo after its prerequisite merged.
There is no hosted tier, no per-seat charge, and nothing to sign up for. You bring the model cost, and that is the whole business model.
Detent runs fully local and is air-gappable. It has no vendor control plane, which means there is no service that can revoke your access, change your pricing, or read your board. The trade is that operating it is your job.
digitaldrywood/detent-orchestration is Detent's own production config. It dispatches the agents that build Detent, and the merged pull requests shown on this site were written by agents it dispatched.
It is copyable as a template for your first project.
Copy the template ↗
One feature. Seven real issues. Four hours. Zero babysitting.
Every line here maps to something checked in, released, or running. Nothing on this site is a projection.
macOS, Linux, and Windows. go install, Homebrew, Winget, Scoop, .deb, .rpm, or copy a single file. No service to stand up.
Runs fully local. There is no Detent vendor control plane and no Detent telemetry — the only outbound traffic is to your tracker and your model provider.
Serialized rebase, CI-watch, and merge — one candidate at a time, so what lands is always green.
Many projects from one host with weights, priority, pause, and fair scheduling.
Token, budget, and rate-limit state live on the dashboard, with caps you configure.
Code defaults to make check plus CI plus automated review. A human approval-label gate is available when the workflow asks for one.
Cross-platform config discovery, a GoReleaser pipeline, and checksum-verified self-update.
You bring the model cost. Nothing else is metered.
"Manage work, not agents" is OpenAI Symphony's phrase. Detent grew out of that open spec and took the idea from spec to a shipped system.
Symphony is an Apache-2.0 SPEC.md plus an Elixir reference implementation that polls a Linear board.
openai/symphony ↗Detent began as an Elixir/OTP implementation adapted from Symphony's Linear target to GitHub Projects v2. It is now a ground-up Go rewrite; that earlier implementation is archived.
Detent's board dispatches agents against its own contracts, and human pull requests go through the same validation gate and the same merge train.
The state machine, the gate model, worktree isolation, and the merge train.
The interaction-model argument, the default choices, and the comparison matrix.
The operator surface: lanes, agent activity, budget and rate-limit state, and the TUI.
Platform-tabbed install commands, requirements, and the first run.
Reviewed recordings of Detent moving real work through checked-in workflows.
The curated operator reference, mirrored from a pinned Detent release.