Skip to content
View rangarajan19's full-sized avatar
0 followers · 1 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report rangarajan19

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
rangarajan19/README.md

Hi, I'm Rangarajan 👋

I build AI agents and the tooling that keeps them safe and measurable.
MCP · tool-use guardrails · agent evaluation · open source

LinkedIn India

🚀 What I'm building

agentkit-mcp

A small MCP-first agent framework where every tool comes from an MCP server, with guardrails at the tool layer: dry-run, human approval, argument rules and tool allow-lists.

  • Own agent loop with tracing and a step limit; tool errors go back to the model so it can recover
  • Works with Gemini and free OpenRouter models, with automatic retry and model fallback
  • Ships two example agents: issue triage (GitHub MCP server plus its own MCP server) and a web research agent
  • Tested with 47 automated tests running in CI, and honest about what is and isn't verified in its README

agentic-api-testing

Chat-driven API testing: a local LLM agent (LangGraph + Ollama) routes requests, while deterministic pytest checks do the actual testing, so the LLM steers and the assertions stay reliable.

🤝 Open-source contributions

Project Merged contribution
HelpCode-ai/anythingmcp #698: added an adapter:new scaffolder so new connector adapters start from a template (closes #585)
agentevals-dev/agentevals #224: CI fix so autofixable ruff violations fail the build instead of passing silently

🧰 Tools I work with

Python MCP LangGraph Ollama Gemini OpenRouter pytest GitHub Actions

🔎 Currently exploring

  • Measuring agent accuracy across models, not just demoing them
  • Agent safety: prompt injection and tool permissions
  • Fixing bugs and improving tooling in projects I use

Feedback and collaboration welcome, reach me on LinkedIn.

Popular repositories Loading

  1. agentic-api-testing agentic-api-testing Public

    Chat-driven API testing: a local LLM agent (LangGraph + Ollama) routes requests, deterministic pytest checks do the testing.

    Python

  2. agentevals agentevals Public

    Forked from agentevals-dev/agentevals

    agentevals is a framework-agnostic evaluations solution based on OpenTelemetry traces

    Python

  3. anythingmcp anythingmcp Public

    Forked from HelpCode-ai/anythingmcp

    🔌 #1 MCP Gateway on GitHub — Connect Claude, ChatGPT, Copilot and any AI client to 200+ ready-to-use connectors, existing MCP servers, and any REST, SOAP, GraphQL or SQL system. Self-hosted, no-cod…

    TypeScript

  4. agentkit-mcp agentkit-mcp Public

    MCP-first agent framework with tool-layer guardrails (dry-run, approval, allow-lists). Gemini free tier. Includes issue triage and research agents.

    Python

  5. rangarajan19 rangarajan19 Public

    Profile README

  6. portfolio portfolio Public

    Portfolio site — Agentic AI Engineer

    JavaScript