| --- |
| title: AgentForge Enterprise |
| emoji: 🧠 |
| colorFrom: indigo |
| colorTo: blue |
| sdk: static |
| app_file: index.html |
| pinned: true |
| license: mit |
| short_description: Governed agentic RAG, MCP and evaluation platform |
| tags: |
| - agentic-ai |
| - langchain |
| - langgraph |
| - mcp |
| - rag |
| - llm |
| - mlops |
| - model-evaluation |
| - responsible-ai |
| - ci-cd |
| --- |
| |
| # AgentForge Enterprise |
|
|
| **Governed Agentic RAG + Tool-Use Platform with LangGraph, LangChain, MCP, Evaluation, Observability and Human Approval** |
|
|
| AgentForge Enterprise is a recruiter-facing demonstration of a production-oriented LLM/agent platform. The public Hugging Face Space is intentionally a free Static Space; the executable FastAPI, LangGraph, LangChain adapter, MCP v2 server, policy controls, tests and Docker implementation live in GitHub. |
|
|
| ## Verified engineering evidence |
|
|
| | Check | Executed result | |
| |---|---| |
| | Dependency installation | Passed | |
| | Ruff quality gate | Passed | |
| | Unit tests | 6/6 passed | |
| | Deterministic contract evaluation | 3/3 passed | |
| | Expected evidence recall in contract cases | 1.0 for all 3 cases | |
| | Approval-policy correctness | Passed for all 3 cases | |
| | Docker image build | Passed | |
| | Dependency audit | Passed after audited build-toolchain hardening | |
|
|
| > **Evidence boundary:** the 3/3 result is a deterministic orchestration/evidence-policy contract evaluation, **not** an LLM quality benchmark. Provider-backed LLM accuracy, groundedness, latency, cost and production reliability are not claimed until separately executed and published. |
|
|
| ## Architecture |
|
|
| ```text |
| User / API |
| ↓ |
| Request validation + policy pre-check |
| ↓ |
| LangGraph supervisor workflow |
| ├── Retrieval / evidence |
| ├── MCP typed tools |
| └── Human approval boundary for side effects |
| ↓ |
| Evidence-backed synthesis |
| ↓ |
| Trace + evaluation artifacts |
| ``` |
|
|
| ## Senior engineering signal |
|
|
| - LangGraph stateful orchestration |
| - LangChain provider abstraction |
| - MCP v2 typed tool interoperability |
| - evidence/provenance contracts |
| - prompt-injection policy checks |
| - allowlisted read/write tool governance |
| - human approval boundary for side effects |
| - reproducible offline evaluation |
| - FastAPI service contract |
| - Docker packaging |
| - CI/CD and dependency auditing |
|
|
| ## Explore |
|
|
| - GitHub source: https://github.com/singhankitsrf/AgentForge-Enterprise-Agentic-RAG |
| - Executed baseline evidence: https://github.com/singhankitsrf/AgentForge-Enterprise-Agentic-RAG/blob/main/evaluation/baseline.json |
| - Hugging Face Space: https://huggingface.co/spaces/singhankit491/agentforge-enterprise |
|
|
