Antelog
Antelog is a runtime safety layer for AI coding agents. It checks actions before execution, blocks destructive or data-leaking actions, and keeps a tamper-evident record so agents can work without accidentally damaging production systems or exposing secrets.
At a Glance
- Individual developers using AI coding agents
- Engineering leads and companies whose agents can reach production systems
- Security, platform and compliance teams needing centralized agent governance
- Teams operating unattended development or infrastructure servers
AI Tools by Antelog
(1)Antelog
Runtime Security for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about Antelog
Latest News
Antelog 0.9.27 released with blocking-only behavior, inline-code sending-check handling and launch-day rule fixes
Antelog release repository reached 27 releases and described the product as early v0.9 software for macOS, Linux and Windows
Public LinkedIn announcement of Antelog's launch as a local security tool that intercepts dangerous AI coding-agent commands
Products & Services
Local runtime guard installed on a developer computer. It checks AI-agent actions before execution, blocks dangerous actions, redacts recognized secrets, and writes a tamper-evident local log.
Free individual plan with local blocking, an online dashboard, and 50,000 hosted AI checks per month per machine plus 14-day online history.
Custom-priced plan for companies that centralizes connected computers, shared rules and company places, approvals, audit logs, CSV export, roles, SSO and self-hosting options.
Market Position
Antelog positions itself as a local, pre-execution safety guard for AI coding tools rather than a general post-deployment observability product: its differentiators are blocking destructive actions on the developer machine, fail-safe behavior, secret redaction, a tamper-evident local log, and team policy/approval controls. SaaSearch lists Squidbrake, LangProtect, Eve Security, Cerbere AG and Hextrap among similar or alternative products.
Leadership
Founders
Yash Sarwaiya
Creator and primary maintainer indicated by the Antelog release repository and its commits. His public profile describes him as a software engineer and University of Florida graduate research assistant, with prior software-engineering work at Bookysh and Profound AI Solutions, plus security research involving Netflix and Lenskart.
Executive Team
Yash Sarwaiya
Creator and primary maintainer
Software engineer and University of Florida graduate research assistant; public profile lists full-stack, distributed-systems, cloud-infrastructure and security experience, including Bookysh and Profound AI Solutions.
Founding Story
The project was created in response to real incidents in which AI coding agents deleted production data and backups. Its initial vision was a local, fail-safe guard that intercepts agent actions before they run, blocks a focused set of destructive and exfiltrating behaviors, and records what happened.
Business Model
Revenue Model
Freemium software: the individual blocking functionality is free, while Team is custom-priced for centralized company controls, approvals, audit history, and enterprise features. Payments are handled through Stripe according to the privacy page.
Pricing Tiers
For one person; local guard, online dashboard, 50,000 AI checks per month per machine, and 14-day online history.
For companies; adds multi-computer management, shared rules and places, approvals, audit log, CSV export, roles, SSO and self-hosting options.
Target Markets
- Individual developers using AI coding agents
- Engineering leads and companies whose agents can reach production systems
- Security, platform and compliance teams needing centralized agent governance
- Teams operating unattended development or infrastructure servers
- Protecting production databases, backups, cloud volumes, buckets and projects from autonomous coding agents
- Preventing AI agents from exposing .env files, private keys, cloud credentials or environment variables
- Stopping prompt-injection-driven uploads, messages, email and pushes to destinations outside an organization's approved places
- Auditing and governing AI-agent activity across developer laptops and unattended servers
- Allowing normal development, testing, reads, builds and deployments while blocking high-impact destructive actions