Highest Evidence AI Agents
Agents with the strongest public evidence depth, weighted toward A/B grades and high HVTrust. Updated from the same generated registry data as the leaderboard, so the list changes with every cron refresh.
16Agents in slice
92Average HVTrust
4Updated in last 14 days
Trust Shape
Activity Signals
| Agent | Category | Grade | Rank | HVTrust | Fresh |
|---|---|---|---|---|---|
| Haystackdeepset-ai/haystack | Agent Frameworks | A | #1 | 96.7 | today |
| Vercel AI SDKvercel/ai | Agent Frameworks | A | #2 | 96.4 | today |
| Codexopenai/codex | Coding Agents | A | #3 | 93.8 | today |
| OmniRoutediegosouzapw/OmniRoute | Agent Skills | A | #1 | 93.5 | today |
| LangGraphlangchain-ai/langgraph | Agent Frameworks | A | #4 | 93.1 | today |
| LiveKit Agentslivekit/agents | Voice & Conversational | A | #5 | 91.5 | today |
| Codebase Memory MCPDeusData/codebase-memory-mcp | MCP Servers | A | #6 | 91.3 | today |
| Strands Agentsstrands-agents/harness-sdk | Agent Frameworks | A | #7 | 91.3 | today |
| Trigger.devtriggerdotdev/trigger.dev | Workflow Platforms | A | #8 | 91.0 | today |
| Clinecline/cline | Coding Agents | A | #9 | 90.9 | today |
| Browser Harnessbrowser-use/browser-harness | Browser & Computer Use | A | #10 | 90.7 | 1d |
| MCP Inspectormodelcontextprotocol/inspector | Protocols & Tool Integration | A | #11 | 90.6 | 1d |
| Playwright MCPmicrosoft/playwright-mcp | MCP Servers | A | #12 | 90.2 | 1d |
| Larksuite Clilarksuite/cli | Agent Skills | A | #2 | 89.9 | today |
| MLflowmlflow/mlflow | Observability & Evaluation | A | #13 | 89.8 | today |
| A2A / Agent2Agent Protocola2aproject/A2A | Protocols & Tool Integration | A | #14 | 89.7 | 1d |
Grade = trust band: A ≥ 80, B ≥ 65, C ≥ 50, D < 50 — methodology