The Candidate Knowledge Graph

The syllabus isn't a flat list — it's a graph with dependencies. An aspirant who studies Art. 368 before Fundamental Rights is studying the amendment mechanism before knowing what it amends. We map those dependencies. Then we map the primary sources onto each node.

A flat list of 400 topics is not a study plan.

The UPSC GS syllabus names domains, not order. The canon books address domains in isolation. No single resource answers the only question that matters: given where I am, what do I study next?

Problem What the canon offers What the CKG offers
Prerequisite blindness Books have internal chapter order; no explicit prereq mapping across the 20+ books an aspirant uses Typed prereq edges: the graph surfaces which nodes are blocked until an upstream concept is solid
PYQ-yield invisibility A few books append PYQ lists; none show yield-per-concept with trend data Each node carries a PYQ-yield count + trend: invest proportionally, not uniformly
Synthesis gap Every book in the canon scores ≤ 2.1 / 5 on cross-topic synthesis (D8) Each node closes on a model Mains answer that draws on the typed edges — the answer demonstrates the cross-topic reasoning, not just the isolated fact
Source ambiguity Coaching books cite "sources" without page references; accuracy erodes across editions Every atom is source-anchored: Constitution article, Economic Survey paragraph, judgment paragraph — verifiable, not claimed

Three layers, built in order.

The CKG is built in three layers. Each layer is independently useful; together they produce a traversal — a daily study path that adapts to what the candidate already knows.

Layer 1 — Canonical terrain

The syllabus as a graph of knowledge units (nodes) with typed edges: prereq / part-of / contrasts-with / applies-to. Each node carries PYQ-yield, difficulty estimate, and centrality (how many downstream nodes it unlocks). Shared, reusable, the expensive asset to build once.

Layer 2 — Content depth

Each node gets a depth ladder (L1–L5) projected from the primary source. L1 is a 90-second plain-language summary. L5 is Mains-ready: source text + the kept/cut/added ledger + a model 15-mark answer with annotated reasoning. Every claim is source-anchored. No coaching notes in this layer.

Layer 3 — Candidate state

A sparse per-candidate mastery rating over the canonical graph. Mastery ratings gate traversal: the CKG surfaces the highest-yield unstudied node that has no blocked prerequisites — the answer to what do I study next? (Layer 3 is in development.)

L1 to L5 — from headline to Mains answer.

Every concept is projected at five depths from the same primary source. An aspirant in Prelims prep uses L1–L2. Mains preparation uses L4–L5. The source is always the same; the register changes.

L1
90-second plain-language summary

What it is, why it matters, one memorable anchor. Written for an aspirant who has never encountered the concept.

L2
Prelims-ready concept

Full definition, key provisions, common MCQ angles, disambiguation from neighbouring concepts. Source reference included.

L3
Mechanism + historical context

How it works procedurally, how it evolved (amendments, landmark cases), what the primary source says in full. Kept/cut/added ledger introduced.

L4
Cross-topic synthesis

How this concept connects to adjacent nodes in the CKG — graph edges rendered as prose. Comparative analysis (e.g., Art. 352 vs 356 vs 360 as one framework). Contested doctrine noted.

L5
Mains-answer ready

Full L4 content plus a model 10/15-mark answer with annotated reasoning: which ethical/governance principle each paragraph activates, why the answer is structured that way. The annotation is as important as the answer itself.

A note on scope: The depth ladder currently covers the static GS syllabus — Constitution, polity, economy fundamentals, history, geography, ethics. Current affairs is deliberately out of scope: the static syllabus is bounded and verifiable; current affairs would require a perpetual newsroom. We'd rather do a finite thing well.

What the graph looks like

This is the Fundamental Rights cluster — 41 nodes, source-anchored, typed edges. A flat yield-list would put Art. 32 (the right to constitutional remedies) at position 1. The graph blocks it: you can't write about Art. 32 without first knowing what a Fundamental Right is, what the original text says in Part III, and how the Basic Structure doctrine constrains its amendment.

# Partial view — FR cluster, prerequisite backbone

Preamble
  └─[prereq]Part III Fundamental Rights    # yield: high · unlocks: 14 nodes
      ├─[prereq]Art 12 — Definition of State  # must precede all FRs
      │   └─[prereq]Art 13 — Laws inconsistent with FRs
      │       └─[prereq]Doctrine of Eclipse
      │           └─[prereq]Doctrine of Severability
      ├─[prereq]Art 14 — Equality before Law
      │   ├─[prereq]Art 15 — Prohibition of Discrimination
      │   └─[prereq]Art 21 — Right to Life
      │       └─[prereq]Maneka Gandhi 1978        # unlocks expanded Art 21
      │           └─[prereq]Art 32 — Right to Constitutional Remedies
      └─[prereq]Art 368 — Amendment Procedure
          └─[prereq]Basic Structure Doctrine       # Kesavananda 1973
              └─[contrasts-with]Parliamentary Sovereignty
        

The flat top-yield list recommends Art. 32 first (9 PYQ appearances). The graph blocks it behind 5 prerequisite nodes — the aspirant who studies Art. 32 without those prerequisites cannot write a Mains answer about it. The graph's #1 recommendation is Art 12 (Definition of State), which unlocks the entire FR cluster.

Try it live — mark what you know →

The diagram above is static. The live version recomputes your study path on every click.

First release: Polity CKG

The FR cluster is built. The full Polity graph is next. Waitlist members get early access and can shape which clusters ship first.

Join the Waitlist