Personal Project
AI Agent
Watcher
When you're running hundreds of agents across multiple projects, visibility becomes critical. Andrew built the AI Agent Watcher — a personal admin panel that provides oversight and governance across every agent in his environment.
Agent Discovery
Scans project directories to find and catalog every configured agent, their permissions, tools, and MCP server connections.
Logs & Audit Trail
Centralized view of what agents have done — tool calls, file edits, command execution — with filtering and search across sessions.
Permissions & Templates
Review and manage agent permission sets, hook configurations, skill assignments, and reusable agent templates across projects.
AI Agent Watcher — built for personal use
Full inventory of every discovered agent across all monitored projects, with status and configuration details.
Activity log showing recent agent actions, tool calls, and file operations across sessions.
Chronological timeline of agent events, showing when agents started, what they did, and how long each session lasted.
Saved agent configurations with their assigned permissions, tools, and model settings.
Creating a new agent configuration with custom permissions, tool access, and behavioral constraints.
Multi-project overview showing planning status, active agents, and progress across workstreams.
Detailed project plan with task breakdowns, dependencies, and agent assignments for a single project.
Catalog of configured hooks with their event types, trigger conditions, and enforcement rules.
Available agent skills with descriptions, showing the reusable capabilities agents can invoke.
Registry of connected MCP servers with their status, capabilities, and tool endpoints.
Marketplace browser for discovering and installing skills, plugins, and MCP server integrations.
Library of reusable agent templates for spinning up pre-configured agents for common workflows.
Detailed view of a single agent showing its capabilities, permissions, connected services, and session history.