Verify Trust Before You Install

Query the OpenA2A Registry for security scans, dependency risk, community consensus, and known advisories on any AI package.

terminal
$ npx ai-trust check server-filesystem
  Resolved: server-filesystem -> @modelcontextprotocol/server-filesystem

  ai-trust v0.2.1

  @modelcontextprotocol/server-filesystem
  ---------------------------------------------
  Trust Level    4 -- Verified
  Publisher      Anthropic
  Last Scanned   2026-03-12
  Advisories     None
  Dependencies   3 (all scanned)
  ---------------------------------------------
  Verdict: safe to install

Installation

Available via Homebrew, npm, or npx. No configuration required.

terminal
# Homebrew (macOS/Linux)
$ brew install opena2a-org/tap/ai-trust

# npm global install
$ npm install -g ai-trust

# Run directly (no install needed)
$ npx ai-trust check express

Commands

Three commands for single-package lookups, dependency file audits, and batch verification.

ai-trust check

Look up trust information for a single package. If not in the registry, scan it locally with HMA.

$ ai-trust check server-filesystem

ai-trust audit

Parse dependency files (.json, .txt) and batch-query all dependencies

$ ai-trust audit package.json --min-trust 3

ai-trust batch

Look up trust verdicts for multiple packages at once, with optional type filtering

$ ai-trust batch express lodash chalk --min-trust 2

Trust Levels

The OpenA2A Registry assigns one of five trust levels to each package based on scan results, publisher verification, and community consensus.

0
Blocked
Package is blocked due to security concerns
1
Warning
Package has known issues or advisories
2
Listed
Package is listed but not yet scanned
3
Scanned
Package has been scanned by HackMyAgent
4
Verified
Package is verified by the publisher

Scan on Demand

Package not in the registry? ai-trust downloads it and runs a HackMyAgent security scan locally. Results are shown immediately and can be contributed as anonymized telemetry to grow the community trust graph.

terminal
# Scan a package not yet in the registry
$ ai-trust check mcp-server-xyz --scan-if-missing

# Scan and contribute results to the community registry
$ ai-trust check mcp-server-xyz --scan-if-missing --contribute

# Force re-scan even if registry data exists
$ ai-trust check server-filesystem --rescan

# Scan missing deps in a dependency audit
$ ai-trust audit package.json --scan-missing --contribute

Community Contribution

Share anonymized scan findings with the OpenA2A Registry to help the community identify unsafe packages. No personal data, no source code -- only check pass/fail results and severity. Your choice is saved and shared across all OpenA2A tools.

Opt-in prompt

On your first scan, ai-trust asks once. Your choice is saved to ~/.opena2a/config.json.

Privacy-first

Only check IDs, pass/fail, and severity are sent. No file paths, descriptions, fix text, or code.

Shared config

Opt-in carries across opena2a-cli, hackmyagent, and ai-trust. Configure once, contribute from any tool.

terminal
# Contribute scan results (non-interactive / CI)
$ ai-trust check chalk --rescan --contribute

# Configure globally
$ opena2a config set contribute true    # opt in
$ opena2a config set contribute false   # opt out

CI/CD Integration

All commands support --json output. Non-zero exit codes when packages are blocked or below the trust threshold.

.github/workflows/trust-gate.yml
- name: Verify AI package trust
  run: npx ai-trust audit package.json --min-trust 3 --json

- name: Check and scan if missing
  run: npx ai-trust check @org/mcp-server --scan-if-missing --json

- name: Audit with local scanning for unknown deps
  run: npx ai-trust audit package.json --scan-missing --contribute

Part of the OpenA2A Ecosystem

ai-trust queries the registry and can also scan locally with HackMyAgent, contributing results back to grow the community trust graph.

HackMyAgent scans

Populates the registry with scan results

ai-trust queries

Checks trust before installation

Registry stores

Trust graph for all AI packages

Check trust in one command

npx ai-trust check server-filesystem