Observal

Active
GitHub Python AGPL-3.0

Description

An observability platform for AI agents that tracks model calls, tool executions, task trajectories, and runtime costs.

Key Features

  • Unified agent registry with one-command install across Claude Code, Cursor, Kiro, and Pi IDEs
  • Full session tracing capturing every prompt, tool call, and response into ClickHouse for analysis
  • AI-powered insight reports analyzing usage patterns with actionable What's Working / Hindering / Quick Wins summaries
  • Package-manager-style agent browser with download counts, ratings, and IDE compatibility badges
  • Admin review queue with version diffs, approval workflows, and leaderboard rankings
  • Enterprise edition with SAML SSO, SCIM provisioning, audit logs, and executive dashboard

Use Cases

💡 Centralize observability for multi-agent coding sessions across a development team
💡 Track model call costs, tool execution failures, and task trajectories for production agents
💡 Compare agent versions via diff views before rolling out updates
💡 Generate executive-ready reports on organizational AI usage and productivity gains
💡 Manage agent distribution and governance with review queues and access control

Tags

agent-monitoring tracing analytics

Categories

📊 Observability

Quick Start

Install the server with `curl -fsSL https://raw.githubusercontent.com/BlazeUp-AI/Observal/main/install-server.sh | bash`. Install the CLI via `pipx install observal-cli`, then run `observal auth login` and `observal doctor --patch` to connect your IDE. Pull agents with `/observal pull security-auditor`.
Visit GitHub

Related Projects