Lead time per task
Time from “accepted scope” to merge-ready PR with approvals done.
Time from “accepted scope” to merge-rea…Evidence
Overview
How Kodus works
Install once. Describe the outcome in plain language. Kodus works in your workspace: it reads and edits the files you open, swaps models mid-conversation, serves a live preview URL, renders line-by-line diffs, and runs security plus design checks before a task can clear - all in one session. Clients cover Mac, Windows, Linux, modern browsers, tablets, and phones so the same thread follows you.
- One workflow, one Definition of Done: Pick tickets that match your reviewer capacity: small scope, observable output, crisp acceptance checklist.
- Turn on guards where it hurts: Auth, billing, public APIs, migrations - anything that cannot ship half-right - gets Review mode, human sign-off, and your usual QA bar.
- Spell out responsibilities: Name who drafts prompts, who approves merges, who owns outages. Ambiguity slows pilots more than tooling ever will.
how-kodus-works-overview.md - operating view
LeadOne workflow, one Definition of Done
ReworkTurn on guards where it hurts
ToolSpell out responsibilities
IncidentProve repeatability twice
Overview
A typical session, end to end
Most people start by opening Kodus beside their repo - or pasting a GitHub URL so the project lands in their workspace automatically. From there one thread owns the whole task.
- One workflow, one Definition of Done: Pick tickets that match your reviewer capacity: small scope, observable output, crisp acceptance checklist.
- Turn on guards where it hurts: Auth, billing, public APIs, migrations - anything that cannot ship half-right - gets Review mode, human sign-off, and your usual QA bar.
- Spell out responsibilities: Name who drafts prompts, who approves merges, who owns outages. Ambiguity slows pilots more than tooling ever will.
- Prove repeatability twice: If the workflow only works once, it isn’t repeatable. Repeat it on a sibling ticket before widening scope.
- Lead time per task: Time from “accepted scope” to merge-ready PR with approvals done.
Active focus:
Overview
Scope
Risk
Output
how-kodus-works-overview.md
Capturing...
// Focus:
A typical session, end to end
1) Pick a narrow slice: One service, one surface, or one class of bug - enough to learn without betting the quarter. 2) Wire review + monitoring: Make sure your existing CI, linters, and human reviewers are in the path before you expand headcount. 3) Compare two weeks of data: Before and after: lead time, rework, incidents. If it doesn’t move a metric you care about, adjust the workflow. 4) Grow by adjacency: Add the next workflow only when the first one is boringly reliable.
1Lead
2Rework
3Tool
4Incident
5Review
What Overview stakeholders get operationally
Rework rate
Share of passes that bounced for missing tests, style, or risky patterns.
Share of passes that bounced for missin…Evidence
Tool stickiness
Sessions per active developer per week in the cohort you onboarded.
Sessions per active developer per week…Evidence
High-signal placements
Overview rollout focus
Most people start by opening Kodus beside their repo - or pasting a GitHub URL so the project lands in their workspace automatically. From there one thread owns the whole task.
Lead time per task
Time from “accepted scope” to merge-ready…Rework rate
Share of passes that bounced for missing…Tool stickiness
Sessions per active developer per week in…Incident or rollback count
Anything that escaped review that you wou…One workflow, one Definitio…
Buyer-ready evidence
Pilot
Review
Evidence
Scale
How it works
Most people start by opening Kodus beside their repo - or pasting a GitHu…
Most people start by opening Kodus beside their repo - or pasting a GitHub URL so the project lands in their workspace automatically. From there one thread owns the whole task.
1) Pick a narrow slice
One service, one surface, or one class of bug - enough to learn without betting the quarter.
2) Wire review + monitoring
Make sure your existing CI, linters, and human reviewers are in the path before you expand headcount.
3) Compare two weeks of data
Before and after: lead time, rework, incidents. If it doesn’t move a metric you care about, adjust the workflow.
FAQ
Questions we hear in pilots
Where does my source code live?
On your machine. Kodus runs models in the cloud, but it does not upload or mount your tree wholesale; it works with what you expose in the session. For air-gapped work, point at a local model and keep traffic on your LAN.
Can I switch models without starting over?
Yes. Swap providers mid-thread from the same chat without re-pasting context.
What if the agent goes off the rails mid-run?
Send a follow-up while it is working, set a budget cap, or stop the run. You can revert from the diff sidebar before anything merges.
How do teams share work?
Invite teammates to the same workspace: shared projects, threads, and @-named skills so the best prompt becomes everyone’s shortcut.
How should we pilot?
Pick one bottlenecked workflow with named reviewers, run two cadence loops, revisit metrics.
Does tooling replace approvals?
No - Kodus complements review, scanners, budgets, and your escalation paths.
Pricing
Pricing
Use the same Kodus plans, tokens, and routing controls across workflows and posture.
Best to Start
Starter
For individual usage.
$20/mo
- 10M tokens / month
- 300 iterations / month
- Standard model routing
- Chat history + projects
- Community support
Team
For small teams.
$100/mo
- 70M tokens / month
- 2,500 iterations / month
- Full routing + Review + Strategy
- Bring your own local model
- Teams (up to 2 members)
- Priority support
- Audit log access
Scale
For larger organizations.
$200/mo
- 300M tokens / month
- 7,500 iterations / month
- Unlimited team members
- All models + custom routing
- Dedicated support channel
- Early access to beta features
- No annual contract
- Tokens reset monthly
- Switch plans anytime
Have invite code? Get Access Now