Research tool only. Company names mentioned are for educational and analytical purposes — not recommendations to buy, sell, or hold any security. Ameya Pimpalgaonkar is not a SEBI registered investment advisor or research analyst. Read full disclaimer →

Claude Cowork Plugin · Free · v3.3.0

Most people read the headline.
We read the value chain.

Turn any technology signal, global event, or market theme into a structured layer-by-layer breakdown — through conversation, not summaries.

Download Plugin See it in action →
6
Framework layers
18
Ready scenarios
All
Your connectors
MIT
Open source
Live decomposition — AI Inference Demand
L0SignalWhy inference is the battleground
L1MechanicsWhat inference actually requires
L2Cause TreeStructural drivers of demand
L3Solution SpaceChip, cloud, model industries
L4Build RequirementsCompute, memory, power, cooling
L5Value Chain ActorsGlobal + India proxies
L6Research LandscapeHow to access this space
L5 → NvidiaAMDAWS 🇮🇳 Tata Elxsi🇮🇳 KPIT
What's new in v3.3.0

Everything in one plugin

India Proxy rebuilt from scratch · Domain Knowledge · Token-efficient decomposition · Consistent design system

Core feature
Animated Value Chain
Layer-by-layer animated diagram after every decomposition. Global companies and India proxies in separate columns.
  • Slides in layer by layer on generation
  • India proxies with distinct visual treatment
  • ⬡ Attribution badge on every output
Rebuilt in v3.3.0
India Proxy Agent
Mandatory 23-step sourcing checklist: NASSCOM Deep Tech, DPIIT, GeM portal, BSE SME / NSE Emerge, 12 VC portfolios, analyst coverage. Map-after-fetch logic — no silent drops.
  • Fetches 12 curated India VC portfolio pages — seed-stage to Series B
  • Sources bar shows exactly what was fetched vs blocked
  • Sub-segment decomposition before every search
Automated monitoring
Watchlist + Trigger Alerts
Weekly / daily digest to Gmail, plus layer-specific triggers that fire only when a precise condition is met.
  • Cadence digest: weekly, daily, fortnightly
  • "Alert me when Layer 4 of EV batteries changes"
  • HIGH / MEDIUM / LOW materiality scoring
Rebuilt in v3.3.0
Scenario Library
18 pre-built topics across 6 categories. Preview all 6 layer hints before committing. One click launches the full decomposition.
  • India · Geopolitics · AI · Cybersecurity · Energy
  • AI drills to Global AI and India AI
  • One click to launch full decomposition
New in v3.3.0
Domain Knowledge
Seed companies the agent won't find via search — pre-coverage, IISc-incubated, DRDO-adjacent. The agent checks your registry before any search.
  • /curiosity-stack:knowledge opens your registry
  • Post-agent "missed a company?" inline form
  • Seeded vs Discovered labels in every output
New in v3.3.0
Token-efficient decomposition
Layers run as plain text conversation. One artifact generated at the end — not 7. ~60% fewer output tokens per session.
  • ~5 topics per session vs ~2 in v3.2.x
  • Visual artifact on demand — say "generate"
  • Progress nudge on every layer so you know it's coming
v3.3.0 improvement

More research per session

Every Claude plan has a session limit. v3.3.0 uses it far more efficiently.

v3.2.x
7×
artifact generations
per decomposition
~2 topics per session
v3.3.0
1×
artifact generation
per decomposition
~5 topics per session

~60% fewer output tokens per full 6-layer decomposition

The difference

Surface reaction vs structured understanding

Most people react to signals. Curiosity Stack turns any signal into a structured, layer-by-layer business insight.

Without Curiosity Stack

Read the headline — form a vague impression, move on

Miss the non-obvious businesses buried in the chain

React when it's already priced in or covered

Can't explain the second and third order effects

With Curiosity Stack

Understand why the signal matters — and why now

Find non-obvious businesses at each layer of the chain

Identify India's specific position in the chain

Walk away with a structured, shareable research brief

The framework

Six layers. One method. Any signal.

One question at a time. You build the understanding yourself. That's the point.

L0
Signal
What's happening and why now?
L1
Mechanics
What is this really at a technical level?
L2
Cause Tree
Root causes — structural vs cyclical?
L3
Solution Space
What industries does each cause create?
L4
Build Requirements
What inputs and infrastructure?
L5
Value Chain Actors
Who does this — global and India?
L6
Research Landscape
How is this space typically accessed?

"The intent is not to reach the end of the process but to go through the process itself."

Scenario library

18 topics ready to decompose

Not sure where to start? Browse pre-built scenarios across six categories. Each launches a full six-layer decomposition with one click inside the plugin.

Install to access all 18 →
Live demos

See a real decomposition

Three real scenarios — exactly as the plugin generates them.

Geopolitics

Middle East conflict → oil supply chain

How escalation travels through the global oil value chain — who wins, who loses, where India sits.

L0
Signal
Strait of Hormuz carries 21% of global oil. Any disruption = immediate price shock.
L1
Mechanics
India imports 85% of crude. $10/barrel spike adds ~₹83,000 crore to the import bill.
L2
Cause Tree
Structural: geographic concentration. Cyclical: current geopolitical tension.

+ L3 Solution Space, L4 Build Requirements, L6 Research Landscape inside the plugin

Value Chain Output — L5 Actors
Global
Saudi AramcoADNOCVitol
Refining
🇮🇳 Reliance🇮🇳 IOC🇮🇳 BPCL
Logistics
Frontline🇮🇳 SCITrafigura
Renewables
🇮🇳 Adani Green🇮🇳 NTPC🇮🇳 Waaree
Non-obvious insight: Renewable companies are paradoxical beneficiaries — each oil spike compresses the payback period on solar and wind.
AI — Global

AI inference demand — the next battleground

Training is won. Inference is what happens when billions of people use AI models every day.

L0
Signal
ChatGPT serves 100M+ users daily. Each query costs ~10x a Google search. Inference is the new infrastructure.
L1
Mechanics
Inference requires different hardware than training — lower latency, higher throughput, different memory profile.
L2
Cause Tree
Structural: AI PMF confirmed. Cyclical: GPU shortage accelerates custom silicon.

+ L3 Solution Space, L4 Build Requirements, L6 Research Landscape inside the plugin

Value Chain Output — L5 Actors
Chips
Nvidia H100AMD MI300Google TPU
Memory
SK HynixMicronSamsung
Packaging
TSMC CoWoS🇮🇳 Tata Elxsi🇮🇳 KPIT
Cooling
VertivSchneider🇮🇳 Airedale
Non-obvious insight: Advanced packaging (OSAT) is the layer most investors miss. TSMC's CoWoS packaging is as much a bottleneck as the GPU itself.
AI — India

Indian IT services in the AI era

AI is automating the work Indian IT services companies sell. Where does opportunity actually sit?

L0
Signal
TCS, Infosys, Wipro all launching AI practices — while traditional services demand softens.
L1
Mechanics
IT services = arbitrage of Indian talent vs Western costs. LLMs now do 30-40% of that work at near-zero marginal cost.
L2
Cause Tree
Structural: LLM capability crossed the routine services threshold. Cyclical: client IT budgets under pressure.

+ L3 Solution Space, L4 Build Requirements, L6 Research Landscape inside the plugin

Value Chain Output — L5 Actors
Legacy IT
🇮🇳 TCS🇮🇳 Infosys🇮🇳 Wipro
AI-native
🇮🇳 Persistent🇮🇳 MphasisAccenture AI
GCC
🇮🇳 Nasscom GCCsJP Morgan India
Data layer
🇮🇳 iMerit🇮🇳 Appen IndiaScale AI
Non-obvious insight: The data annotation layer is India's emerging advantage — same talent pool that built IT services can build the human-in-the-loop layer AI depends on.
Watchlist + alerts

Set it. Forget it. Get alerted.

Add any topic to your watchlist. The plugin monitors every layer of the value chain on your schedule — and fires an immediate alert the moment something material changes.

Step 1 — Add a topic
After any decomposition say "add to watchlist". Or open the watchlist and add directly. Pick your cadence — daily, weekly, or fortnightly.
Step 2 — Set trigger conditions
Tell it exactly what to watch for. "Alert me when a new player enters L4." "Flag any funding round in this space." As specific or broad as you want.
Step 3 — Receive alerts to Gmail
Scheduled digests land in Gmail on your chosen day. Trigger alerts fire immediately — marked HIGH priority — the moment a condition is met.

Two monitoring modes

Tracker mode — scheduled digest
A comprehensive summary of everything that changed across all layers of your watchlist topic. Delivered to Gmail on your schedule.
"Add EV batteries to my watchlist, weekly digest"
→ Added. Digest every Sunday to Gmail.
Trigger mode — instant alert
A precise condition at a specific layer. Fires immediately when met — separate HIGH priority email, regardless of your digest schedule.
"Alert me when a new company enters L4 of green hydrogen"
→ Trigger set. Fires immediately when detected.

What a digest email looks like

WEEKLY DIGEST · ENTERPRISE CYBERSECURITY Sunday 08:00
HIGH
L4 trigger fired — New funding round: Cyera raised $300M Series D. Enters L4 as a data security posture management player.
MED
L3 shift — SIEM market consolidation continues. Palo Alto acquiring IBM QRadar assets signals L3 consolidation at the platform layer.
LOW
L6 access update — CrowdStrike added to MSCI India index. India proxy access via listed route now available.
3 signals · 1 trigger fired · Next digest: Sunday ⬡ Curiosity Stack

Manage everything in one place

Watchlist manager
2 topics · 4 triggers active
+ Add topic
Enterprise cybersecurity demand
3 triggers Weekly
New funding or product company emerges
SIEM or SOC market share shift at L3
India proxy listing or funding event
AI inference demand
1 trigger Weekly
New inference chip announced or sampled
Gmail connected · Next digest: Sunday 08:00
Schedule →
How to get started
After any decomposition say "add to watchlist" — the plugin sets it up instantly with a UI, no YAML involved. Or open the manager directly:
/curiosity-stack:watchlist
Trigger alert flow: When a trigger condition is met, an immediate HIGH priority email fires to your Gmail — separate from your regular digest. The watchlist UI shows which trigger fired and when.
In action

Watch it work

Real plugin sessions. See the Socratic conversation, layer cards, and final value chain output.

Video coming soon

Core feature
Full 6-Layer Decomposition — Semiconductor Packaging
Signal to animated Value Chain output. ~10 min.

Video coming soon

Autonomous agent
India Proxy Agent — Finding Indian Equivalents of Nvidia
Live autonomous search with validated shortlist. ~6 min.

Video coming soon

Monitoring
Watchlist Tracker — Setting a Layer 4 Trigger Alert
Configure EV battery monitoring with a layer trigger. ~5 min.

Video coming soon

New in v3.3.0
Scenario Library — Browse and Launch in One Click
18 ready-made topics, one-click decomposition launch. ~4 min.
Who uses it

Built for two kinds of thinkers

📊
Investors and researchers
Retail investors · Analysts · Fund researchers
Find non-obvious businesses before they're covered, before they're priced in. From vague signal to named companies at each layer — India proxies flagged specifically.
🏢
Enterprise professionals
Strategy · Business development · Product · Consulting
Understand where AI and technology are actually going — not just the headline, but the business implications, the value chain, the India angle.

Free. Open source.
Install in 2 minutes.

Download from GitHub and install directly in Claude Cowork. MIT licensed — inspect, fork, and customise freely.

v3.3.0MIT LicenseClaude Cowork52 filesall your connected tools
Download v3.3.0 from GitHub
Announcement post Privacy policy Browse source
1
Download the plugin
Click the button → download curiosity-stack-v3.3.0.zip
2
Open Claude Cowork
Claude Desktop → Cowork tab → Customize → Browse Plugins
3
Upload the zip
Upload custom plugin → select zip → installs immediately
4
Run setup
Type /curiosity-stack:setup — 2 minutes — then describe any topic
Regulatory Disclaimer & Important Notice

Curiosity Stack is built and operated by Ameya Pimpalgaonkar (@Finstor85). Neither the plugin nor its operator is a registered Research Analyst under SEBI (Research Analysts) Regulations, 2014. This website and all content herein, including outputs generated by the Curiosity Stack plugin, has been prepared solely for informational and educational purposes. Nothing contained on this website or produced by the Curiosity Stack plugin constitutes or should be construed as investment advice, a research report, a recommendation, or a solicitation to buy or sell any security, financial instrument, or asset class.

Not a Trade Recommendation
The views, analysis, and value chain maps expressed on this website and generated by the plugin are based on publicly available information. They should not be relied upon as the basis for any investment or trading decision. Readers must conduct their own independent due diligence before acting on any information contained herein.
No SEBI Registration
This content is not produced by a SEBI-registered Research Analyst, Investment Adviser, or Portfolio Manager. No regulatory oversight applies to the opinions expressed here. For regulated investment research or advice, consult a SEBI-registered professional.
Investment Risk
Investments in securities markets are subject to market risks. Past performance is not indicative of future results. Companies discussed may be subject to risks not covered in this content. Stock prices are volatile and can move materially against any position. Capital loss is possible.
Company References
Companies named at each layer of any value chain are identified for analytical and educational purposes only. Their inclusion does not imply a recommendation to buy, sell, or hold their securities. The author may or may not hold positions in named companies. Such holdings represent personal decisions, not managed portfolio recommendations.
Data & Sources
All data referenced is sourced from publicly available information including company filings, exchange disclosures, SEBI publications, RBI data, and financial databases. While care has been taken for accuracy, no warranty is given as to the completeness or accuracy of any data.
AI-Assisted Research
All plugin outputs are generated using Curiosity Stack — a structured research framework running on Claude (Anthropic). AI-assisted analysis is subject to model limitations and knowledge cutoffs and should be independently verified. This does not alter the non-advisory nature of this content.
Not SEBI registered  ·  Not investment advice  ·  For informational and educational purposes only  ·  curiositystack.app
⬡ Curiosity Stack