Technology layer
Agentic
Software that plans, calls tools, and runs multi-step work on behalf of a company — not raw model weights.
The Agentic layer is where products stop being “a chat box over an LLM” and start being *operators*: they break goals into steps, bind tools and memory, and hand judgment back to humans at policy gates.
In this catalog it holds the largest share of surveyed technologies — frameworks, RAG stacks, gateways, evals, guardrails, coding and voice agents, and the protocols that let agents talk to tools safely.
Use this layer when you are assembling an agentic OS spine (observe → understand → propose → gate → execute). Pair it with Infrastructure for durable stores and LLMs for serving and fine-tuning.
See also
Technologies
- ABMGitHub
- AG2GitHub
- Agent EvaluationGitHub
- Agent GatewayGitHub
- Agent2Agent ProtocolGitHub
- AgentScopeGitHub
- AgnoGitHub
- AiderGitHub
- Arcade MCPGitHub
- Arize PhoenixGitHub
- Atomic AgentsGitHub
- AutogenGitHub
- AutoGPTGitHub
- BeeAI FrameworkGitHub
- Browser UseGitHub
- BurrGitHub
- CAMELGitHub
- CedarGitHub
- Comet OptikGitHub
- ComposioGitHub
- ContinueGitHub
- Crawl4AIGitHub
- Crew AIGitHub
- Dapr AgentsGitHub
- DaytonaGitHub
- DeepEvalGitHub
- deer-flowGitHub
- DifyGitHub
- DoclingGitHub
- DSPyGitHub
- e2bGitHub
- Earned Autonomy ProtocolGitHub
- EnvoyGitHub
- FastMCPGitHub
- FirecrawlGitHub
- Google ADKGitHub
- Guardrails AIGitHub
- GuidanceGitHub
- HAL: The Holistic Agent LeaderboardGitHub
- HatchetGitHub
- HaystackGitHub
- HeadroomGitHub
- HeliconeGitHub
- higressGitHub
- InngestGitHub
- InstructorGitHub
- Jina ReaderGitHub
- JulepGitHub
- Jupyter Enterprise GatewayGitHub
- KeycloakGitHub
- kgatewayGitHub
- KongGitHub
- KubernetesGitHub
- LangChainGitHub
- LangflowGitHub
- LangFuseGitHub
- LangGraphGitHub
- LangMemGitHub
- LangroidGitHub
- LangSmithGitHub
- LettaGitHub
- LiteLLMGitHub
- LiveKit AgentsGitHub
- LlamaIndexGitHub
- LLM GuardGitHub
- LMQLGitHub
- markitdownGitHub
- MastraGitHub
- MCP Gateway RegistryGitHub
- MCP RegistryGitHub
- MCP Servers (reference)GitHub
- mem0GitHub
- MetaGPTGitHub
- Microsoft Agent FrameworkGitHub
- MinerUGitHub
- MirascopeGitHub
- Model Context ProtocolGitHub
- NeMo GuardrailsGitHub
- Open InterpreterGitHub
- Open Policy AgentGitHub
- OpenHandsGitHub
- OpenLLMetryGitHub
- OutlinesGitHub
- PaddleOCRGitHub
- PipecatGitHub
- PM4PyGitHub
- PocketFlowGitHub
- PortKeyGitHub
- PromptfooGitHub
- Pydantic AIGitHub
- RagasGitHub
- RAGFlowGitHub
- RigGitHub
- ScraplingGitHub
- Selenium Grid 4GitHub
- Semantic KernelGitHub
- SmolagentsGitHub
- StrandsGitHub
- SupergatewayGitHub
- SWE-agentGitHub
- TemporalGitHub
- TruLensGitHub
- txtaiGitHub
- UnstructuredGitHub
- VocodeGitHub
- VoltAgentGitHub
- VoyagerGitHub
- W&B WeaveGitHub
- WhisperGitHub
- ZepGitHub
Features
- 01-company-model-memory-os
- 02-execution-os
- 03-ai-workforce-os
- 04-unified-company-cockpit-os
- 05-agent-infrastructure-os
- 06-vertical-company-os
- ABM span mapping
- Agent Building Platform
- Agent Identity & Interoperability
- Agent Lifecycle Management
- Agent Supply Chain Integrity
- APIs, SDKs & MCP Extensibility
- Approval calibration
- Architecture
- Assistant ↔ Expert
- Assistants
- Authoring
- Automation / operations
- Autonomy Levels
- Bring Your Own Model
- Browser & Computer Use
- CaaS Platform
- Central nervous system
- Channels
- Citations
- Coding agents
- Company observation
- Context engineering
- Control Flow
- Cost, Latency & Operational Economics
- Customer support
- Dashboards
- Data & Context
- Decision lineage
- Decision-package human surfaces
- Deployment & Data Residency
- Design agents
- E-commerce
- Ecosystem Master Synthesis
- Education
- Emergency Control / Kill Switch
- Eval harnesses
- Evaluator / critic
- Finance & Accounting
- Four planes, copy/reject, 10 invariants, full index
- Gap Matrix
- Governance
- Governance
- Healthcare
- HR & Recruiting
- Human & Policy
- Human Oversight & Guardrails
- Humans for exceptions/judgment
- Importance consolidation
- Independent post-condition checks
- Insurance
- Integrations & Tool Calling
- Intelligence
- Intervention pre-assessment
- Knowledge Grounding & RAG
- Learning & Adaptation
- Legal
- Master Synthesis v2
- MCP Tools
- Memory & State Persistence
- Memory architecture
- Model Flexibility & Routing
- Multi-Agent Collaboration
- Observability & Auditability
- OCED case events
- Open Source
- Perception & Multimodal Sensing
- Persistent company model ≠ agent memory
- Prebuilt Agents, Templates & Packs
- Process mining
- Pulse/events ≠ long Case resolution
- Real Estate
- Research agents
- Reuse, Deploy & Govern
- Runtime
- Sales / GTM
- Sandboxing & Isolation
- Security
- Security, Identity & Governance
- See ≠ may-act
- Self-coding
- Self-Host
- Self-monitoring
- Single-Purpose Agent
- Skill registry
- Sleep
- Sleep orchestration
- Supply Chain & Logistics
- Temporal Coherence / Long-Horizon Consistency
- Test, Debug & Evaluate
- Testing, Debugging & Optimization
- Tools & Actions
- Trace maturity gate
- Triggering
- Triggers & Channel Coverage
- Unknown → preserve → inspect → map
- Verify business effect, not API 200
- Voice Agent Platform
- Workflow Orchestration
Stacks
- Agentic → Agent Protocols
- Agentic → Agentic Gateways
- Agentic → Code Sandbox
- Agentic → Coding Agents
- Agentic → Evaluations
- Agentic → Frameworks
- Agentic → Guardrails
- Agentic → IAM
- Agentic → LLM Gateway
- Agentic → Memory
- Agentic → Observability
- Agentic → RAG
- Agentic → Runtime
- Agentic → Search & Web Access
- Agentic → Structured Output
- Agentic → Tool Integration
- Agentic → Voice Agents
- Agentic → Workflow Orchestration
- durable-gated-execution
- mcp-tool-plane
- observe-eval-loop
- rag-with-citations
Document management
GitHub in this term
Primary repositories linked from member packs and devices.
- ABMa-rebmann/ABM
- AG2ag2ai/ag2
- Agent Evaluationawslabs/agent-evaluation
- Agent Gatewayagentgateway/agentgateway
- Agent2Agent Protocola2aproject/A2A
- AgentScopeagentscope-ai/agentscope
- Agnoagno-agi/agno
- AiderAider-AI/aider
- Arcade MCPArcadeAI/arcade-mcp
- Arize PhoenixArize-ai/phoenix
- Atomic AgentsEigenwise/atomic-agents
- Autogenmicrosoft/autogen
- AutoGPTSignificant-Gravitas/AutoGPT
- BeeAI Frameworki-am-bee/beeai-framework
- Browser Usebrowser-use/browser-use
- Burrapache/burr
- CAMELcamel-ai/camel
- Cedarcedar-policy/cedar
- Comet Optikcomet-ml/opik
- ComposioComposioHQ/composio
- Continuecontinuedev/continue
- Crawl4AIunclecode/crawl4ai
- Crew AIcrewAIInc/crewAI
- Dapr Agentsdapr/dapr-agents
- Daytonadaytonaio/daytona
- DeepEvalconfident-ai/deepeval
- deer-flowbytedance/deer-flow
- Difylanggenius/dify
- Doclingdocling-project/docling
- DSPystanfordnlp/dspy
- e2be2b-dev/E2B
- Earned Autonomy ProtocolLatamlinkLLC/earned-autonomy-protocol
- Envoytheagentrouter/agent-router
- FastMCPPrefectHQ/fastmcp
- Firecrawlfirecrawl/firecrawl
- Google ADKgoogle/adk-python
- Guardrails AIguardrails-ai/guardrails
- Guidanceguidance-ai/guidance
- HAL: The Holistic Agent Leaderboardprinceton-pli/hal-harness
- Hatchethatchet-dev/hatchet