Run your own process across many agents at once.
This page covers the argument, the two default choices that define Detent's footprint, and the comparison matrix, including the rows where Detent loses.
Where the intelligence lives.
Autonomy-first tools put the intelligence in the agent and ask you to supervise it. Detent puts the intelligence in your spec and supplies the discipline. Same task, two interaction models.
- It starts as a prompt. The agent picks a plan.
- You read the plan and inspect partial edits.
- It misses the migration. You redirect.
- It skips the tests. You redirect again.
- You are the supervision loop.
- It starts as an issue naming the storage change, CLI behavior, migration, rollback, and tests.
- A runner executes that contract in an isolated worktree.
- It produces a reviewable Change Request.
- It does not land until your gates clear.
- Your process runs. You read one change.
Detent scales the judgment of people who already hold a high bar. It applies your rules the way you would, every time and in parallel, and every step stays visible on the board.
And who should close the tab.
The person who runs delivery
An engineering lead, a producer, an operations owner. You already have review standards, approval rules, and an order things happen in. Your bottleneck is throughput on well-specified work without lowering the bar.
You want the state machine, the gate model, serialized landing, and the isolation story. You will read a detent.yaml before you read a testimonial. There is no testimonial on this site.
The solo builder with five projects
One person, several repositories or pipelines, a queue of work that is clear but unwritten. You want parallelism without babysitting five chat sessions.
Start on the Free plan with no card, enroll the laptop you already have, and bill agent work against the Codex or Claude Code subscription you already pay for.
The platform team
Evaluating agent work for an organization. Your questions are governance questions.
Agents on machines you control, monthly budgets and capacity limits, runner routing per project, review decisions on the Hub, and a self-hosted Hub when the work cannot leave your network.
People who want an autonomous assistant to talk to. People looking for a copilot inside an editor. People who want a vendor to run the agents on its own machines. Detent is none of those, and it is better for everyone if that is clear here rather than three pages into the docs.
Two choices define the footprint. Both are deliberate.
The Hub coordinates. Your machines do the work.
The Hub holds the board, the issues, the conversation history, the review decisions, and the policy each project runs under. Runners are machines you enroll, and they keep your checkout, your agent sign-ins, and your toolchain.
Nothing that touches your code runs on the Hub. That is why Detent Cloud can be the default without holding your checkout or your credentials, and why the same Hub can run on your own host when it must.
Your subscriptions pay for the model.
Runners drive agents through their own CLIs, Codex through codex app-server, Claude Code through the claude CLI, using the sign-in already on the machine. Detent stores no model credentials.
Model and reasoning effort are set per organization and project in the Hub. Pi and local models through Ollama are supported too. A Detent Cloud plan pays for hosting and concurrent work, never for tokens.
Including the rows Detent loses.
This matrix is copied from docs/comparison.md in the repository. Where a competitor is better, the table says so.
| Capability | Detent | Copilot agent | Cursor | Hermes | OpenClaw | Hyperagent |
|---|---|---|---|---|---|---|
| Self-hosted, no vendor control plane | ✓ yes self-hosted Hub | ✗ no | ✗ no | ✓ yes | ✓ yes | ✗ no hosted |
| Agents run on your own machines | ✓ yes enrolled runners | ✗ no | ◐ partial self-hosted workers | ✓ yes | ✓ yes | ✗ no cloud |
| Board/tracker-native (issue→change) | ✓ yes native tracker, optional GitHub | ✓ yes GH Issues | ✗ no | ✗ no | ✗ no | ✗ no own workspace |
| Deterministic gated landing | ✓ yes | ✗ no | ✗ no | ✗ no | ✗ no | ✗ no |
| Budget / cost caps | ✓ yes | ◐ partial | ◐ partial | ✗ no | ✗ no | ✓ yes hosted controls |
| Multi-project | ✓ yes | ✓ yes | ✓ yes | ✗ no | ✗ no | ◐ partial workspace-scoped |
| Runner fleet routing and capacity | ✓ yes | ✗ no | ✗ no | ✗ no | ✗ no | ◐ partial hosted agent controls |
| Model choice, BYO subscription | ✓ yes Codex and Claude Code | ✓ yes vendor-managed | ✓ yes vendor-managed | ✓ yes | ✓ yes | ✗ no cloud-managed |
| Local skills / workflows (your e2e etc.) | ✓ yes | ✗ no | ◐ partial | ✓ yes | ✓ yes | ✓ yes hosted skills/knowledge |
| Multi-channel triggers | ◐ partial tracker, conversations, MCP | ◐ partial GitHub | ◐ partial IDE/cloud tasks | ✓ yes messaging gateway | ✓ yes local gateway | ✓ yes Slack, schedules, webhooks, email, Telegram, Live Mode |
| Source available | ✓ yes FSL-1.1-ALv2 | ✗ no | ✗ no | ✓ yes MIT | ✓ yes MIT | ✗ no closed-source |
| Single static binary | ✓ yes | — not applicable SaaS | — not applicable SaaS | ✗ no gateway | ✗ no gateway | — not applicable SaaS |
✓ yes · ◐ partial · ✗ no · — not applicable. Last verified October 7, 2026; check vendor pricing and models before relying on them.
Single-binary Go orchestrator: a Hub (Detent Cloud or self-hosted) holds the board and policy, and enrolled runners execute issues as gated changes.
GitHub's paid issue- or prompt-to-branch-and-PR agent.
IDE-first agent product with cloud and background agents, automations, and optional self-hosted workers.
Nous's MIT personal assistant with memory, skills, model providers, and a messaging gateway.
MIT personal assistant centered on a local gateway and cross-channel automation.
Airtable's closed-source, cloud-hosted agent platform for persistent agents with identity, tools, and knowledge.
Copilot and Cursor have closed much of the “runs near my code” gap and win zero-install inside their own platforms. They do not give you a board-native release runtime under your control. Hermes, OpenClaw, and Hyperagent reach more channels than Detent does.
The questions people actually ask.
Will it ship garbage?
Not without passing the gates you wrote: your gate command on the finished change, an automated validator, a security audit, or a person in Human Review. Then the runner that holds the project lands one change at a time, and a change that cannot land goes to Blocked with its reason.
What if the agent goes off the rails?
It works on its own branch in its own worktree, and it cannot land anything; only the runner lands, after your gates clear. Moving the issue out of an active lane revokes the worker and stops its process. A worktree is collision isolation, not an OS sandbox, so the guarantee covers what lands. It does not limit what the process can touch.
Is it only for code?
No. Lanes, roles, and the gate command are yours. A video pipeline or an approval workflow is a board with different lanes and a different definition of done.
Does my code go to Detent Cloud?
Your checkout stays on your runners, and nothing that touches your code runs on the Hub. Change Requests do carry their diff to the Hub so people can review them, along with issues, comments, and run history. If that is too much, self-host the Hub for free.
Is this a toy?
Judge it the way you would judge any repository: version history, a CI badge, a release channel, installer smoke tests on Ubuntu and Windows against real release assets, and the fact that it builds itself on Detent Cloud.
What is the catch?
You bring the model cost, and you write the contracts. Detent will not invent your acceptance criteria, and it is not trying to. A vague issue produces vague work faster than before.
Start free on Detent Cloud. Paid plans add concurrent work and hosted artifacts.
The Hub and runners, the state machine, the gate model, worktree isolation, and serialized landing.
Install a runner or your own Hub: platform-tabbed commands, requirements, and the first run.
The operator surface: lanes, agent activity, budgets and capacity, and MCP access.
Reviewed recordings of Detent moving real work through checked-in workflows.
The curated operator reference, mirrored from a pinned Detent release.
FSL-1.1-ALv2, the repository, releases, contributing, and how Detent builds itself.