AI Guardian MCP server

Read from the package source without running it. The installed server may list more.

• anomaly_report

  [READ] Rollup: shadow models, digest drift, high-risk prompts, blocked count.

• guarded_generate

  [WRITE][risk=medium] Scan + policy-gate a prompt, record it, then run if allowed.

• list_models

  [READ] Installed models, each annotated with the allow/deny policy verdict.

• mainwrite action

  Run the MCP server over stdio.

• model_details

  [READ] License / parameters / capabilities for one model.

• model_provenance

  [READ] Compare each installed model's digest against its pin; flag drift.

• observe_chat

  [WRITE][risk=medium] Scan + policy-gate a chat exchange, record it, then run if allowed.

• pin_model_digest

  [WRITE][risk=medium] Pin a model's expected provenance digest (drift detection).

• policy_view

  [READ] The current model allow/deny policy + provenance digest pins.

• proxy_guidance

  [READ] How to run the transparent capture proxy, and what it does NOT guarantee.

• pull_model

  [WRITE][risk=medium] Pull a model — refused if it violates the deny/allow policy.

• remove_modelwrite action

  [WRITE][risk=high] Delete a local model. Destructive — pass dry_run=True to preview.

• running_models

  [READ] Currently loaded models: VRAM footprint + residency expiry.

• scan_prompt

  [READ] Pure content scan of a text for secrets / PII / code / jailbreak.

• server_status

  [READ] Ollama reachability + version.

• set_model_allowlist

  [WRITE][risk=medium] Replace the model allowlist (glob patterns; empty = allow-all).

• set_model_denylist

  [WRITE][risk=medium] Replace the model denylist (deny patterns always win).

• undo_apply

  [WRITE][risk=medium] Apply a recorded undo by dispatching its inverse tool.

• undo_list

  [READ] List recorded, not-yet-applied undo tokens (most recent first).

• unload_model

  [WRITE][risk=medium] Evict a model from VRAM (keep_alive:0).

• usage_events

  [READ] Query the observed-usage log (route-through prompts + their findings).

• vram_usage

  [READ] Total VRAM used by loaded models; flag over-budget.

• Code scan55 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitylocal package, no credentials required12/15
• Maintenancelast push 3 days ago15/15
• Maintainer identityregistry namespace matches repository owner7/10

AI Guardian: common questions

Is AI Guardian MCP server safe?

Yes, by our scan: it is graded A (91/100). Read the AI Guardian safety report

How do I install AI Guardian?

It runs on your machine. Copy the Claude Code, Claude Desktop or Cursor config from the install section.

Does AI Guardian need an API key?

Not as far as the registry entry and our scan can tell: no credentials are declared or required.

Is AI Guardian maintained?

The last commit was 3 days ago (2026-09-16). The latest release is v0.11.3.

What can I use instead of AI Guardian?

Servers from other publishers that do the same job: Agent-Native Analytics MCP server, Maginary MCP server and Protect MCP server. Compare all AI Guardian alternatives.