AI agent governance buyers are trying to put limits around what agents can access, where data can go, how tool use is approved, and whether activity can be audited. The June 2026 ToolVitals data for AI Agent Governance Tools covers open/source-visible projects with scores for health, shipping activity, openness, licensing, GitHub stars, and an overall ToolVitals score.
This ranking uses open/source-visible tools only, ordered first by ToolVitals score and then by GitHub stars as a secondary popularity signal.
Rankings
| Rank | Tool | Openness | License | Health | Shipping | GitHub Stars | Score | Status |
|---|---|---|---|---|---|---|---|---|
| 1 | mcpproxy-go | OSI-approved OSS | MIT | 89 | 98 | 272 | 94 | 🟢 Excellent |
| 2 | Pipelock | OSI-approved OSS | Apache-2.0 | 85 | 95 | 738 | 92 | 🟢 Excellent |
| 3 | Panguard AI | OSI-approved OSS | MIT | 84 | 95 | 50 | 92 | 🟢 Excellent |
| 4 | ArifOS | OSI-approved OSS | AGPL-3.0 | 83 | 83 | 48 | 87 | 🟢 Excellent |
| 5 | Multicorn Shield | OSI-approved OSS | MIT | 79 | 76 | 3 | 83 | 🟢 Excellent |
| 6 | Idun Agent Platform | OSI-approved OSS | GPL-3.0 | 74 | 66 | 193 | 77 | 🟢 Good |
| 7 | Clawmoat | OSI-approved OSS | MIT | 67 | 61 | 40 | 74 | 🟢 Good |
| 8 | RelayPlane Proxy | OSI-approved OSS | MIT | 67 | 49 | 185 | 70 | 🟢 Good |
| 9 | ShieldPilot | OSI-approved OSS | Apache-2.0 | 57 | 14 | 1 | 50 | 🟡 Fair |
Top 3 Highlights
mcpproxy-go ranks first with a 94 ToolVitals score, the highest score in the set. Its 98 shipping score is also the strongest listed signal, and it carries an 89 health score, 272 GitHub stars, OSI-approved OSS openness, and an MIT license_label. Its description, “Supercharge AI Agents, Safely,” is broad, but the score profile is the clearest reason it leads this June 2026 list.
Pipelock ranks second with a 92 ToolVitals score and the largest GitHub star count in the ranking at 738. It is described as an AI agent firewall for MCP security and egress control, which puts it directly in the governance lane. It is labeled OSI-approved OSS with an Apache-2.0 license_label, and its 95 shipping score suggests strong recent project activity in the supplied data.
Panguard AI also has a 92 ToolVitals score, but ranks behind Pipelock because its GitHub stars are lower at 50. Its description is more security-platform oriented, covering AI agents, skill audits, and threat monitoring. The project is labeled OSI-approved OSS with an MIT license_label, and its 84 health score plus 95 shipping score keep it in the Excellent group.
Want to see the full health, shipping, openness, and evidence details for every tool in AI Agent Governance? Browse all AI Agent Governance Tools →