Browser MCP Defense rule blocks external Codex agent tool calls before execution
HankYeomans · x · 2026-10-18
HankYeomans demonstrated a side-by-side test of an external agent (Codex) calling tools via browser, with and without a "Browser MCP Defense" rule:
- Without the rule: the agent's searchFlights (SFO→JFK) call completed, returning FLT-101 at $480 — notably dated March 20, 2026, a past date, i.e. suspicious data.
- With the rule: the same searchFlights call was blocked pre-execution with "Tool call blocked by policy before execution: Tool call matched a Wave policy rule," and no results were returned.
Shows a practical way to add a policy layer over MCP tool calls to intercept suspicious calls before they run.
More from coding & agent
- Building a multi-agent shared knowledge base with two GitHub repos — and one memory pitfall — gregbarbosa · 2026-10-18
- Fully agentic MV pipeline: Claude + MiniMax H3 renders 28 scenes with 0ms sync drift — Cheap_Credit_3957 · 2026-10-18
- Dev builds seamless laptop/web/mobile dev env with Claude Code self-hosted runners + Tailscale — davidcrawshaw · 2026-10-18
- SOC Prime launches LogTotal: rule-first log analysis with AI interpretation at the end — cyb3rops · 2026-10-18
- Thorsten Ball: A 'wrong' code push matters far less now that AI makes changes cheap — sloppenheimer · 2026-10-18
- Every trains a Slack agent to edit like its editor-in-chief using a custom skill — every · 2026-10-18