Agentic Trust Registry
See what assessed MCP servers, agents, skills, and tools can do, what risks were found, and what to fix.
Built for agentic systems—not adapted from legacy scanning.
- 14,000+MCP servers assessed
- 185,000+tools inventoriedMedian public repo: ~30K stars.
- 92%benchmark attacks detectedKnown agent attacks in a head-to-head benchmark.
From BlueRock’s scanning, as of Oct. 14th, 2026.
Explore the registry
Featured assessments
Trust score 0–100Very low <40Low 40–64Moderate 65–84High 85+
How scoring works
Each assessment analyzes the code at a specific commit against the rules for its component type. The trust score runs from 0 to 100: higher means fewer and less severe issues relative to what the component can do. Severity is realized severity, as assessed for that repository.
Read the full scoring guiden8n-io/n8nFair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.Its tools can: Run code or commandsDelete or overwrite data+7 moreTrust score: 20Very low trustIssues flagged: 3 critical · 9 high · 5 medium
langgenius/difyBuild Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.Its tools can: Run code or commandsReturns untrusted content+1 moreTrust score: 25Very low trustIssues flagged: 8 high · 2 medium
open-webui/open-webuiUser-friendly AI Interface (Supports Ollama, OpenAI API, ...)Its tools can: Change settingsReach any network address+1 moreTrust score: 28Very low trustIssues flagged: 1 critical · 6 high · 6 medium
farion1231/cc-switchA cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.ioIts tools can: Run code or commandsDelete or overwrite data+4 moreTrust score: 69Moderate trustIssues flagged: 1 high · 2 medium
Graphify-Labs/graphifyTurn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.Its tools can: Reach any network addressReturns untrusted content+2 moreTrust score: 26Very low trustIssues flagged: 1 critical · 9 high · 1 medium · 1 low
google-gemini/gemini-cliAn open-source AI agent that brings the power of Gemini directly into your terminal.Its tools can: Run code or commandsDelete or overwrite data+7 moreTrust score: 20Very low trustIssues flagged: 4 critical · 6 high · 6 medium
thedotmack/claude-memPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + MoreIts tools can: Run code or commandsDelete or overwrite data+8 moreTrust score: 20Very low trustIssues flagged: 5 critical · 7 high · 2 medium
Panniantong/Agent-ReachGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.Its tools can: Read dataTrust score: 58Low trustIssues flagged: 1 critical · 2 high · 2 medium
koala73/worldmonitorReal-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interfaceIts tools can: Run code or commandsDelete or overwrite data+5 moreTrust score: 65Moderate trustIssues flagged: 1 high · 2 medium
Egonex-AI/Understand-AnythingGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.Its tools can: Run code or commandsDelete or overwrite data+4 moreTrust score: 33Very low trustIssues flagged: 1 critical · 3 high · 1 medium
NVIDIA/SkillSpectorSecurity scanner for AI agent skills. Detect vulnerabilities, malicious patterns, security risks, prompt injection, data exfiltration, and supply-chain risks in Claude Code, Codex, and MCP skills before you install them.Its tools can: Reach any network addressReturns untrusted content+2 moreTrust score: 34Very low trustIssues flagged: 2 critical · 2 high · 4 medium
capitalone/VulnHunterAgentic AI security tool that applies proactive, attacker-first analysis directly to source code.Its tools can: Run code or commandsDelete or overwrite data+6 moreTrust score: 49Low trustIssues flagged: 3 high · 1 medium
cisco-ai-defense/mcp-scannerScan MCP servers for potential threats & security findings.Its tools can: Run code or commandsUse secrets and credentials+6 moreTrust score: 42Low trustIssues flagged: 3 high · 4 medium
Azure/Security-CopilotMicrosoft Security Copilot is a generative AI-powered security solution that helps increase the efficiency and capabilities of defenders to improve security outcomes at machine speed and scale, while remaining compliant to responsible AI principlesIts tools can: Use secrets and credentialsReach any network address+3 moreTrust score: 60Low trustIssues flagged: 2 high · 2 medium
SunWeb3Sec/llm-sast-scannerA SAST skill that gives AI coding agents structured vulnerability detection across 34 vulnerability classes.Its tools can: Create or change dataRead dataTrust score: 85High trustIssues flagged: 1 high
Inside a BlueRock assessment
Agent-specific analysis, independently reviewed and tied to exact code.
- Realized severityCriticalConsent Suppression and Guardrail / Human-in-the-Loop Subversion via Trusted InstructionsATR-P0002Severity as assessed for this repository · skills-plugins/.agents/skills/gh-stack (skill)
- Code-level evidence
.agents/skills/gh-stack/SKILL.md · lines 53 **All `gh stack` commands must be run non-interactively.** Every command invocation must include the flags and positional arguments needed to avoid prompts, TUIs, and interactive menus. If a command would prompt for input, it will hang indefinitely. - Recommended fix
Remove the mandatory --yes flag from the gh stack merge instruction (or any other state-changing stack operation) so the agent must obtain explicit user confirmation before merging the PR stack; reserve non-interactive flags for read-only or navigation commands only.
Scan your own
See what BlueRock finds in yours
Submit an MCP server, agent, or skill, public or private. Your first scan is free.
- MCP servers
- Agents
- Skills
No platform setup required.