Library
The front door of this site is deliberately short. This page is the rest of it: every concept page, comparison, spec, and data set, in one place. Nothing here has moved; it is just no longer in the top menu.
Concepts
- What is an Agentic Control Plane?
- What is an Agent Harness?
- What is Runtime Authorization?
- What is Tool Call Economics?
- What is an Agent Delegation Chain?
- What is an MCP Control Plane?
- What is an Agentic Data Plane?
- What is an Agent Firewall?
- Harness Control Patterns
- Rules: tool, tier, user
- Agent Identity
- Agent-to-Agent Control and delegation chains
- ADCS — the open Agent Delegation Chain Specification
Comparisons and positioning
- Why you need an Agentic Control Plane
- What an Agentic Control Plane is not
- The three planes: identity, control, observability
- Agent identity vs. agent control
- What a gateway can actually enforce
- ACP vs ACP: Agent Client Protocol or Agentic Control Plane?
Proof and data
- Caught in the Wild — real interceptions from our own sessions
- The Data — every first-party number we have published
- AgentGovBench — the benchmark scorecard and all 48 scenarios
- The Coding Agent Tool Surface Index
- Harness Coverage Matrix
- Harness controls, compared and MCP server controls, compared
Guides and use cases
- Getting started and what the installer does, file by file
- Every way to set up ACP
- Use cases — the incidents ACP stops, catches, and proves
- Agent personas — ready-made rule sets
- ACP for coding agents
- ACP for teams
- Cost tracking and the Cost X-ray
- AI agent cost monitoring, the guide
- The Agent Optimization Manual
- Reference architecture
- Frameworks — guided starters
- Implementation guides: Okta, Auth0, Entra ID, LangChain, Vercel AI SDK, PII detection, SOC 2 audit trails
Series
- Stop Your Agent From…
- Recreated — real incidents, rebuilt and prevented
- MCP Security
- AgentGovBench
- Governed Agent Recipes
- Three-minute quick starts per framework
Everything else
- FAQ · About · Community · Trust & Security · Blog
Running coding agents? ACP records each tool call its hook sees and makes the dangerous shell commands (a force-push, destructive SQL, curl | sh) wait for you. One command, free up to 5 agents:
macOS · Linux · WSL
curl -fsSL https://agenticcontrolplane.com/install.sh | bashWindows · PowerShell
irm https://agenticcontrolplane.com/install.ps1 | iex