# Aegis

> GitHub branch protection rules, but for AI agents. Aegis intercepts every tool call your AI coding agents make and enforces Allow, Deny, Rewrite, or Require Approval policies before anything touches GitHub.

## Docs

- [Decisions](https://docs.runaegis.co/concepts/decisions.md): Learn how Aegis assigns ALLOW, DENY, REWRITE, or REQUIRE_APPROVAL to every MCP tool call, and how blast radius influences enforcement.
- [Policies](https://docs.runaegis.co/concepts/policies.md): Understand Aegis policies: what they are, how the 10 built-in rules work, how to enable or disable them, and how policy state is stored.
- [Rooms](https://docs.runaegis.co/concepts/rooms.md): Learn how Aegis rooms work: what they are, how to create and join them, role permissions, tool policies, and how to generate the integration URL.
- [Sessions](https://docs.runaegis.co/concepts/sessions.md): Learn how Aegis sessions group agent tool calls, what session fields mean, how to inspect individual actions, and how approvals interact with sessions.
- [GitHub setup](https://docs.runaegis.co/configuration/github-setup.md): Provide your GitHub credentials and personal access token so Aegis can inspect your repositories, evaluate agent actions, and manage pull requests.
- [Notifications](https://docs.runaegis.co/configuration/notifications.md): Choose where Aegis sends alerts (email, webhook, or Slack) and control exactly which governance events trigger a notification for your account.
- [Repository permissions](https://docs.runaegis.co/configuration/repository-permissions.md): Control which repositories your AI agents can access and whether they can act autonomously or must wait for human approval before writing to each repo.
- [Webhooks](https://docs.runaegis.co/configuration/webhooks.md): Register an HTTPS endpoint to receive signed JSON payloads from Aegis whenever a governance event fires: approvals, denials, rewrites, anomalies, and more.
- [Approvals](https://docs.runaegis.co/features/approvals.md): Learn how Aegis pauses agent write actions that require human sign-off, where to find pending requests, and how to approve or deny them from the dashboard.
- [Audit trail](https://docs.runaegis.co/features/audit-trail.md): Understand what the Aegis audit trail records, how to browse and filter events by date range, how to export records, and how long data is retained.
- [Freeze windows](https://docs.runaegis.co/features/freeze-windows.md): Create, edit, and delete Aegis freeze windows to automatically block all agent write actions during scheduled periods like release nights or change freezes.
- [Token usage](https://docs.runaegis.co/features/token-usage.md): Track input and output token consumption per agent session in Aegis, understand free tier limits, and configure notification thresholds to avoid overages.
- [How it works](https://docs.runaegis.co/how-it-works.md): Learn how Aegis intercepts MCP tool calls, evaluates them against your policies, and enforces Allow, Deny, Rewrite, or Require Approval decisions.
- [Claude Code](https://docs.runaegis.co/integrations/claude-code.md): Add Aegis governance to Claude Code by dropping an MCP config snippet into ~/.claude/mcp.json or .claude/mcp.json. Every tool call is logged and policy-checked.
- [Cursor](https://docs.runaegis.co/integrations/cursor.md): Add Aegis governance to Cursor by adding an MCP server entry to ~/.cursor/mcp.json. Agent actions are logged and policy-checked from the first tool call.
- [Integrations](https://docs.runaegis.co/integrations/overview.md): Connect Aegis to Claude Code, Cursor, or VS Code Copilot via MCP over SSE. No code changes required. Just paste a config snippet and restart your tool.
- [VS Code Copilot](https://docs.runaegis.co/integrations/vscode-copilot.md): Govern VS Code Copilot with Aegis by adding a .vscode/mcp.json config to your project. Every agent tool call is intercepted, logged, and policy-checked.
- [What is Aegis?](https://docs.runaegis.co/introduction.md): Aegis intercepts every tool call your AI coding agents make and enforces Allow, Deny, Rewrite, or Require Approval policies before anything touches GitHub.
- [Quickstart](https://docs.runaegis.co/quickstart.md): Create your Aegis account, connect GitHub, set per-repo permissions, and wire up your first AI coding agent via MCP SSE in five minutes.