Technical feasibility

Built on interfaces that exist today

The first version needs no future capability. Agent vendors already let external programs run before and after each tool call.

Terminal demo

Hook integration

Claude Code hooks (PreToolUse and PostToolUse) call the local ffax command, which writes each event to the local log. A PreToolUse hook can return an allow, deny or ask decision with a reason shown to the agent, so blocking happens before anything executes.

MCP proxy

MCP is an open protocol. FFaX sits between the agent and MCP servers and inspects each tools/call.

Cross-vendor coverage

Cursor and other agents also offer hooks and MCP support. FFaX maps them to one event format, vendor by vendor.

One-line setup

.claude/settings.json
{ "hooks": {
  "PreToolUse":  [{ "matcher": "*", "hooks": [{ "type": "command", "command": "ffax hook pre"  }] }],
  "PostToolUse": [{ "matcher": "*", "hooks": [{ "type": "command", "command": "ffax hook post" }] }]
} }
Workflow

Five-step workflow

From install to console in five commands

ffax install · ffax connect · ffax policy apply default · ffax decoy plant · ffax console. Each step is one command on the developer's machine; nothing runs in the cloud.

Open source and local-first

The client is open source and auditable. Data stays on the developer's machine by default, and teams can self-host the console.

Why deterministic matters

Decoy alerts don't rely on a model's guess. Normal work never touches a decoy, so a touch is a reliable signal with no false positives.

Design partner program

Install the demo for free and shape the product

Engineering teams using AI coding agents can run the six-week demo at no cost: open-source client, an action log that never leaves your machine, three default policies, decoys and a one-page console.