The complete platform for AI-powered engineering
One agent. Every developer. Zero friction.
A single background agent captures Claude Code, Cursor and Copilot activity and git output on each machine and syncs it to your organization. Nothing about how developers work changes.
3 KB
agent footprint
3 min
prompt sync interval
0
workflow changes needed
Every developer
Laptops & workstations
- macOS, Windows and Linux
- Claude Code, Cursor and Copilot sessions
- Git repositories you point it at
Runs in the background
CloudByte agent
- 3 KB footprint, one install line
- Hooks plus a background sync job
- Pushes new data every 3 minutes
Your organization
acme.cloudbyte.ai
- Dashboards, alerts and REST API
- Data isolated per organization
- SSO, RBAC and audit log
Your keys, your data, your stack
Connect GitHub and Google SSO, keep working alongside the tools your team already uses, and on the Enterprise plan bring your own Anthropic key and database.
Bring your own Anthropic key
On the Enterprise plan: your key, your billing account. Keys are encrypted at rest with AES-256-GCM and only decrypted per request.
Bring your own database
Store sessions, spend and commit data in your own database, so it stays inside your infrastructure and under your retention rules.
GitHub OAuth
Connect GitHub for repository and pull request context, without webhooks or branch protection changes.
Google SSO
Domain-restricted sign-in with OAuth 2.0 PKCE, plus passwordless magic links for everyone else.
- Claude Code
- Cursor
- GitHub Copilot
- GitHub
- GitLab
- Bitbucket
- Google Workspace
- AWS
Exactly what the agent collects
A short, fixed list, so your security team can review it in one sitting.
Prompts
The text of each prompt sent to Claude Code, Cursor or Copilot.
Observations
Tool calls and results a session produced.
Sessions
Start, end, duration and project of each session.
Summaries
The summary written at the end of each session.
Git commits
SHA, message, author and diff stats, plus the diff for summaries (kept 5 days).
Pull requests
Title, status and review timing for linked PRs.
Token usage
Input, output and cached tokens per model.
CLAUDE.md compliance
Whether each project has a CLAUDE.md and sessions follow it.
Agent heartbeats
Version, platform and status every 3 minutes.
Skills usage
Which shared skills ran, and when they were overridden.
Nothing beyond this list — no keylogging, no file contents, no environment variables.
Built for teams, secured for enterprise
Every control a security team checks for, built in from day one — not bolted on at renewal time.
Google SSO & Magic Link
OAuth 2.0 PKCE with domain-restricted access per org, plus passwordless magic-link email auth. One-time tokens expire in 5 minutes.
Role-Based Access Control
Permissions assigned per role — Super Admin, Org Admin, Reviewer, Standard. Hierarchical and server-enforced on every request, zero client trust.
- Super Admin
- Org Admin
- Reviewer
- Standard
Multi-Project Tenancy
Every project has its own isolated data scope and access boundary. Cross-project data never leaks, even for admins switching between teams within the same organization.
Bring Your Own Anthropic Key
On the Enterprise plan: your Anthropic API key, your billing account. Managed or self-managed tiers. AES-256-GCM encryption at rest.
API-First Architecture
Every dashboard feature is also a REST endpoint. OpenAPI 3 spec shipped, Bearer-token auth, 40+ endpoints covered.
Audit & Compliance
Every AI action logged with user, org, prompt, and token spend. CLAUDE.md compliance tracking per developer per project.
Sensitive Prompt Alerts
Automatically flags prompts containing secrets, credentials, or PII before they leave the developer machine. Real-time alerts to admins. Full audit trail for compliance review.
Secret detected in prompt
Severity: High- Pattern
- Cloud access key (AKIA…)
- Action
- Blocked before upload
- Developer · project
- raj@acme.dev · infra-terraform
- Notified
- 2 org admins · 09:12
- OAuth 2.0 PKCE
- AES-256-GCM at rest
- OpenAPI 3 spec
- Server-side RBAC
- SOC 2 Type II (in progress)
From zero to a working dashboard in under 10 minutes
No infrastructure to run, no repositories to configure, and no new commands for your developers to learn.
- 1
Sign up
30 secCreate your org with Google SSO or an email address. Pick your subdomain (e.g. acme.cloudbyte.ai). No credit card, no infrastructure.
- 2
Install the agent
2 minOne command per developer laptop — macOS, Windows, and Linux supported. Registers hooks, starts the background sync job, and performs an initial data flush.
- 3
See insights
instantPrompts appear on your dashboard within 3 minutes of the first AI session. Commits within 10 minutes. Real team data — not sample data.
$curl -fsSL https://acme.cloudbyte.ai/install.sh | bash
Detected macOS 15 (arm64)
[ok]Claude Code, Cursor and Copilot detected
[ok]Git capture enabled for 6 repositories
[ok]Background sync scheduled (every 3 min)
[ok]Initial flush: 214 prompts, 38 commits
CloudByte agent is online. Open acme.cloudbyte.ai to see your dashboard.
$
Product questions, answered
How the agent works, what it collects, how it connects to your stack and how long your data is kept.
One lightweight background agent of about 3 KB. It installs with a single command on macOS, Linux or Windows in about a minute, runs quietly in the background and needs no change to how developers write code or use their AI tools.
It reads the local session data of Claude Code, Cursor and GitHub Copilot without modifying it and picks up commits from the repositories you point it at. New prompts, sessions and commits sync to your organization every 3 minutes, so the dashboard stays close to real time.
No. It only reads files that already exist and sends small batches of metadata on a short interval. It never sits between the developer and their AI tool, so prompts and responses are not delayed.
It sends prompts and tool observations, session details and summaries, token usage per model, commit metadata with diff stats, and agent heartbeats. Commit diffs used for summaries are kept for only 5 days. It never collects source code files, keystrokes, environment variables, screen recordings or browser activity.
Yes, on the Enterprise plan. With your own Anthropic key, AI spend is billed straight to your Anthropic account and the key is encrypted at rest with AES-256-GCM. You can also bring your own database, so captured sessions, spend and commit data stay inside your infrastructure and under your retention rules.
Claude Code, Cursor and GitHub Copilot for AI activity, GitHub through OAuth for repository and pull request context, and Google SSO for sign-in, alongside GitLab, Bitbucket and AWS. Connecting GitHub needs no webhooks and no changes to branch protection.
Each agent sends a heartbeat every 3 minutes. Agent Health shows every machine as Online, Partial, Initiated, Stale or Offline in one fleet view, so broken installs and machines that stopped syncing are easy to find.
Free covers the activity feed, agent health and basic AI insights for up to 5 developers. Starter adds spend and ROI analytics, commit and pull request summaries, the skills library and GitHub. Team adds unlimited history, the knowledge base, prompt enhancement, AI security, shadow AI detection and Google SSO. Enterprise adds bring your own key, SAML, self-hosting and custom retention.
It depends on your plan: 7 days on Free, 90 days on Starter and unlimited history on Team. Enterprise retention is agreed in your contract.
Partially. On the Enterprise plan you bring your own AI key and your own database, so your keys, AI spend and captured data stay in your own accounts and infrastructure. The CloudByte AI dashboard and backend run on our side, so you always get the latest features, fixes and security updates without managing upgrades yourself.
Put the whole platform on your team today.
Install once per laptop and your dashboard fills with real sessions, spend and commits within minutes.
- Free for up to 5 developers
- 10-minute setup
- Your Anthropic key, your spend (Enterprise)