Live · 9 May 2026 · Speaking at 11th DOT NET DAY by MDCA Ahmedabad

Recap up · AIMED workshop "The Agentic Operating System" · 2 May 2026

Daily agentic AI updates in the WhatsApp community — join from /card

RAGFlow v0.25.6 adds an autonomous browser component, a one-line @tool decorator, and dataset-level retrieval

Field note: the agent reliability bug is almost always a tool contract, not the model

Supply-chain worm harvests Claude Code credentials: poisoned Nx Console VS Code extension breaches GitHub, hits OpenAI and Mistral dev machines

Anthropic opens Claude Security in public beta as Project Glasswing reports 10,000+ critical vulnerabilities found

NVIDIA ships Verified Agent Skills: scanned, signed skill packages with machine-readable risk cards

Cheap open-weight models now ~60% of OpenRouter usage: DeepSeek, Kimi and Zhipu reshape the cost floor

MCP locks its 2026 spec release candidate: the protocol goes stateless ahead of a July 28 final

Cursor 3.5 brings Automations into the Agents Window and adds multi-repo and no-repo cloud agents

Note · Your agents aren't broken. Your tools are.

Note · Agentic AI has an execution problem.

Note · Your agent has no memory. That is the problem.

Live · 9 May 2026 · Speaking at 11th DOT NET DAY by MDCA Ahmedabad

Recap up · AIMED workshop "The Agentic Operating System" · 2 May 2026

Daily agentic AI updates in the WhatsApp community — join from /card

RAGFlow v0.25.6 adds an autonomous browser component, a one-line @tool decorator, and dataset-level retrieval

Field note: the agent reliability bug is almost always a tool contract, not the model

Supply-chain worm harvests Claude Code credentials: poisoned Nx Console VS Code extension breaches GitHub, hits OpenAI and Mistral dev machines

Anthropic opens Claude Security in public beta as Project Glasswing reports 10,000+ critical vulnerabilities found

NVIDIA ships Verified Agent Skills: scanned, signed skill packages with machine-readable risk cards

Cheap open-weight models now ~60% of OpenRouter usage: DeepSeek, Kimi and Zhipu reshape the cost floor

MCP locks its 2026 spec release candidate: the protocol goes stateless ahead of a July 28 final

Cursor 3.5 brings Automations into the Agents Window and adds multi-repo and no-repo cloud agents

Note · Your agents aren't broken. Your tools are.

Note · Agentic AI has an execution problem.

Note · Your agent has no memory. That is the problem.

Jigar Joshi

From AI-assisted coding to AI Systems Engineering.

Jigar Joshi trains development teams and consults with IT companies to build production-grade agentic AI systems using Claude API, MCP, and multi-agent patterns.

Explore Training Book a Call

Trained teams at

Bytes Technolab

Metizsoft

+ Growing

agent_loop.py

<span style="color:#5C7A5A"># ReAct Agent Loop</span>
<span style="color:#7EB8F7">def</span> <span style="color:#FDB94A">run_agent</span>(task):
  <span style="color:#5C7A5A"># Think</span>
  plan = llm.reason(task)
  
  <span style="color:#7EB8F7">while not</span> plan.is_complete():
    <span style="color:#5C7A5A"># Act</span>
    result = tools.execute(plan.next)
    
    <span style="color:#5C7A5A"># Observe</span>
    plan.update(result)
    
  <span style="color:#7EB8F7">return</span> plan.result

Agent running · Step 3 of 5

Claude API

MCP · RAG · Multi-Agent

16 hrs

Per Training Program

80%

Hands-on Every Session

Day 1

Start Building Agents

100%

Working Deliverables

Daily agentic AI updates in WhatsApp

What I am building, what shipped today, and Q&A with builders.

What Jigar Does

Three ways to work together

Corporate Training

Team Training Programs

Intensive 8-day programs that take development teams from AI-assisted coding to building production multi-agent systems. Every session produces working code.

Agentic AI Consulting

For companies that want Jigar to design and build their first agentic system. Architecture, implementation, and handoff to your team.

Daily AI Updates

What shipped in Agentic AI today - curated, condensed, and delivered with implementation notes for builders.

The Shift

What training actually changes

AI-assisted coding

AI Systems Engineering

Using Cursor for autocomplete

Building autonomous agent systems

Single-prompt interactions

Multi-step agent workflows

Manual API integrations

MCP-standardized tool access

No production awareness

Observability, guardrails & deployment

See full curriculum

Topic Pillars

Browse the knowledge graph

Each pillar is an authority hub aggregating every blog post, note, ship-news update, and talk on that topic. Start with the pillar; follow the cluster links for depth.

Agentic AI

Designing, building, and shipping production agents.

Model Context Protocol (MCP)

The open protocol that gives agents tools.

Multi-Agent Systems

Orchestrating many agents without losing the plot.

Claude API

Building production agents on Anthropic's Claude.

AI Observability

Tracing, eval, and telemetry for production agents.

AI Engineering

The discipline of shipping AI systems, not demos.

Enterprise AI Automation

Operational agents for IT services and enterprise teams.

Latest Writing

From the blog

ProductionMay 27 · 12 min

Your agent's supply chain is the attack surface now

A poisoned VS Code extension spent eighteen minutes on the marketplace and walked off with Claude Code credentials and MCP configs. The model was never the target. Your agent's supply chain is: the extensions, skills, MCP servers, tool definitions, and keys it is allowed to touch. Here is how I harden all four layers, and the checklist I run on every deployment.

MCPMay 26 · 11 min

MCP just went stateless: what the 2026 spec release candidate changes for your servers

The biggest revision of MCP since 1.0 locked as a release candidate on May 21. The protocol goes stateless, extensions move out of the core, and authorization finally speaks OAuth properly. Most of your servers keep working. Here is what actually changes, what breaks, and the migration I would run in the ten weeks before the final spec lands.

Tool DesignMay 25 · 11 min

Your agents aren't broken, your tools are: three questions to ask before you build one

When an agent misbehaves, almost everyone reaches for the prompt or the model. The fault is usually further down, in a tool that does too much, lies when it fails, or buries the answer in a wall of raw data. An AI tool is not a function. It is a contract the model has to trust. Here are the three questions I run before writing a single line of any tool.

Follow the Work

Daily Agentic AI content on every platform

Ready to build?

Your team is one training away from
shipping agent systems.

Customised to your stack. Hands-on from day 1. Working deliverables every session.

Book a discovery call See curriculum