Doberman Review 2026: Runtime guardrails that gate your AI coding agent
Doberman is runtime guardrails for AI coding agents. It sits on the execution path as a transparent MCP proxy or host hook and intercepts every input, output, and tool call with a PASS / AUTH / BLOCK verdict. It fails closed.
What Doberman Does
Doberman is the βwatchdogβ for AI coding agents. Positioned on the runtime execution path as a transparent MCP proxy or host hook, it intercepts every input, output, and tool call before execution and returns one of three verdicts: PASS (zero friction), AUTH (pause for human approval), or BLOCK (never runs). On any error or uncertainty it fails closed β the action is denied.
Key Features
- Fail-closed: uncertain or errored actions are denied, never silently run
- Three verdicts β PASS, AUTH (with repeat-action quick-confirm), BLOCK
- MCP proxy or native host hook for Claude Code, Codex, Cursor, Claude Desktop
- Audit logs, telemetry toggle, and strictness dials (Light to Paranoid)
- Self-protecting: agents cannot rewrite Dobermanβs own config or state
Who Should Use Doberman
Developers running autonomous coding agents who fear an rm -rf-style disaster, and security/platform teams needing policy enforcement and audit trails for agent fleets.
Pros and Cons
Pros
- Strong default posture β fail closed on doubt
- Flexible integration (MCP or host hooks)
- Auditable with adjustable strictness
Cons
- Setup and policy tuning add overhead
- Heuristic false positives on obfuscated or nested commands
- Narrow audience β mainly security-conscious operators
Pricing
Free and open source under Apache-2.0.
FAQ
Which agents does it support?
Claude Code (hooks), Codex CLI (PreToolUse), Claude Desktop/Cursor (MCP proxy), OpenClaw (native), and any MCP-compatible agent.
What happens if approval times out?
A timed-out approval is recorded as timeout and denied β fail-closed by design.
Explore the best AI Coding tools
Related Articles
Aether Review 2026: Turn Your AI Subscription Into a Fleet of Devboxes
Aether (runaether.dev) turns the AI subscription you already pay for into parallel cloud devboxes where an agent streams every command, opens a PR, and another agent reviews and fixes it until the code holds up. We review the loop, the receipts, and the pricing.
agent-run Review 2026: Run Coding Agents in a Tiny Sandbox That Catches Mistakes Before They Spread
In-depth review of agent-run β a sub-1MB standalone binary that sandboxes coding agents (Claude Code, Codex, OpenCode, pi) inside a Bubblewrap container. Host filesystem is read-only by default. Built to catch agent mistakes, not malware.
Best AI Agent Tools in 2026: From Coding Assistants to Autonomous Workers
Complete guide to AI agent tools in 2026 β Claude Code, Codex, Cursor, Manus, and more. Which agents actually deliver on the promise of autonomous work?
Faultsense Review 2026: The expect() Without the Page
Faultsense is a zero-dependency browser agent that runs end-to-end assertions against real user sessions in production. We review how fs-* attributes work, RUM-style testing, and who should adopt it.
Subscribe to the 9bests weekly β get the full list free
Hand-picked AI tool reviews and updates every week. Subscribe to receive this full list + 7 more quick-reference sheets (writing / image / video / audio / chat models / data / API cost).
Subscribe free & get it βIndependent reviews β ratings aren't influenced by vendor payments Β· double opt-in Β· unsubscribe anytime