Pluggable runtime.
Not a Claude wrapper. Not a Codex wrapper. The agent is a swappable runtime, so you change models or vendors without changing platforms.
The control plane for coding agents.
Your agent fixes failing pipelines.
Bring your own agent. No lock-in.
Claude Code
Codex
Z.ai
Cursor soon 
soon
Gemini soon
Ollama soon 
opencode soon + whatever ships next
soon
Gemini soon
Ollama soon
opencode soon + whatever ships next How it works
Events in. Adapters route. Results out.
Everything is an event. Every integration is an adapter. The engine takes the event, runs the right agent in isolation, and reports back to where it came from.
Events come in
An MR comment, a failed pipeline, a webhook, a chat message. Platform adapters (GitHub, GitLab, Telegram, Slack) normalize each one into a single task event.
The right agent runs, isolated
The engine picks the agent adapter (Claude Code, Codex, Z.ai) and spins an ephemeral runner with the repo checked out and your credential injected at run time. Nothing persists on disk.
It calls back where it came from
The result becomes events going the other way. The MR is opened or updated, the pipeline retried, the comment posted, the chat replied to. One control plane, one audit trail.
Why vemlor
The control layer for agent work.
Different sources.
Most tools are GitHub-first with GitLab bolted on. vemlor treats both as first-class.
Team-ready by default.
Auth, roles, audit, allowlists, per-workspace tokens. Priced so one dev can adopt it before the team does.
Self-hostable from day one.
Run it on your infra with your keys and your data. No "enterprise tier" gate. Single Docker image or Helm chart, same product.
Integrations
Plugs into the tools you already use.
It listens where your team already works, and reports back where they look.
GitHub
GitLab
Telegram
Slacksoon
Sentrysoon
Linearsoon
Jirasoon
Notionsoon
The economics
Flat price. Your tokens stay yours.
We don't sell models, credits, or inference. The model cost is yours, paid through the subscription you already have. Our job is the platform.
You pay vemlor for
- Runners and webhooks
- Auth, RBAC, audit log
- Self-host or VPC, day one
- Updates, support, the roadmap
You never pay vemlor for
- Per-token charges
- Credit packs
- Per-seat surcharges
- Overage on a busy week
Two separate relationships: your model provider for inference, vemlor for the platform around it.
Pricing
Start free. Stay flat.
From $0 to $299/mo flat. Enterprise self-host from $30k/yr. One unit: the autofix. No tokens, no credits, no surprise bills.
Free
$0 forever
Try the platform with no card.
Hobbyists, OSS, and solo trials.
- 1 repo
- 30 autofixes / mo
- Bring your own agent
- No team workspace
- No audit log or RBAC
- No priority support
Most popular
Team
$99$79 / month, flat
$99/mo billed monthly
≈ 4 hrs of senior dev time. Pays for itself in week 1.
Small teams (3-20 eng). Whole team in, no seat counting.
- 10 repos · unlimited users
- 1,000 autofixes / mo
- Shared workspace, RBAC, audit log
- All integrations
Growth
$299$239 / month, flat
$299/mo billed monthly
≈ 10 hrs of senior dev time. Replace a part-time CI babysitter.
Mid-market (20-100 eng). More repos, more pipeline.
- 50 repos · unlimited users
- 5,000 autofixes / mo
- Priority webhooks, policy controls
- Dedicated success channel
Enterprise
SSO/SAML, SCIM, self-host or VPC, audit export, DPA, SLA. From $30k/yr.
"vemlor pays for itself the first week. We dropped our flaky-CI babysitting from ~6 hrs/week to almost zero, and the price is the same every month no matter how many fixes ship."
Two platforms. Six agents. One queue.
GitHub and GitLab as first-class peers. Claude Code, Codex, Z.ai live; Cursor, Aider, Gemini, Ollama and opencode landing. Every run reports back to one control plane.
Join the waitlist.
Signups are paused while we onboard the first cohort. Drop your email and we'll invite you the moment a slot opens.
Already invited? Sign in →