Search

AI model: AI45Research/AgentDoG-FG-Llama3.1-8B (text-classification)

Read-only static scanner for AI-agent codebases: finds dangerous action sinks, proves reachability, attributes guard coverage, and rates findings honestly (reachable-verified vs install-liability).

PikaKit — AI Operating System. 51 skills, 18 workflows, strict TypeScript.

AI model: mradermacher/AgentDoG-FG-Qwen3-4B-i1-GGUF

SidClaw governance integration for LangChain — policy evaluation, human approval, and audit trails for AI agent tools

Agentic SOC analyst for Splunk — Claude investigates via the Splunk MCP Server; a deterministic 3-layer validator verifies every finding against real data. An anti-hallucination trust layer for agentic SOC.

AI model: golf-mcp/golf-prompt-guard

证据约束的代码诊断工作台：确定性静态分析拥有源码事实，LLM 只提假设，resolver/validator 决定能否绑定回真实操作——证据不足保持 unknown/abstained，从架构抑制幻觉。已在真实开源项目发现 bug（stunner #89 已被上游确认修复）。Python。

Public Preview — AgentMesh Runtime: Execution supervisor for multi-agent sessions with privilege rings, saga orchestration, and audit trails

LLM-powered decision layer for the S2N web vulnerability scanner

Node SDK for checking AI agent tool calls with Relay.

Enterprise tool for discovering AI Agents running within an organization

Cloudflare OS is Cloudflare’s open source AI productivity environment for building company-specific agent workspaces. It combines an agent chat UI, generated workspace apps called gadgets, Workers-based sandboxing, and Gatekeepers that mediate access to external systems with narrow permissions, logging, and human approval.

Windows computer-use MCP server (26 tools) with a human-in-the-loop approval gate: physical-input-only dialogs, signed guard switches, deny/launch lists, hash-chained audit log. 70+ unit tests, green CI.

Pi extension for web search, page extraction, Firecrawl scraping/crawling, and Crawl4AI headless browser crawling.

Sanitize LLM outputs before HTML, SQL, shell, or markdown sinks. Python port of @mukundakatta/llm-output-sanitizer.

AI tool hook adapter for punkgo-kernel — every tool call gets a cryptographic receipt

Framework-agnostic core for AI Kit - streaming, agents, state management, and LLM primitives

Universal AIGP (AI Governance Protocol) client — consent-based runtime AI governance

Give AI coding agents case-by-case sudo access via GUI approval, with SSH-key-encrypted password storage