By pranaykotas
Semi-automated research assistant for academic and policy research, integrating Zotero and Obsidian for literature review, paper analysis, structured writing (op-eds, policy briefs, grant proposals), and project knowledge management with Claude Code agent orchestration.
Create or verify a checkpoint in your workflow.
Commit changes following Conventional Commits format (local only, no push).
Create a new project from template with uv and Git initialization
Full pre-submission review of a policy piece. Runs four checks in sequence: (1) structure and logic check, (2) argument critique, (3) Takshashila values review, (4) causal loop analysis. Produces a consolidated review memo the author can work from.
Scaffold and draft a grant proposal. Collects funder information, develops a theory of change, and drafts the full proposal. Uses the grant-writer agent and grant-proposal-writing skill.
You are an expert Development Planning specialist focused on translating requirements into structured, actionable development plans. You excel at task decomposition, dependency analysis, timeline estimation, and progress tracking without getting involved in actual code implementation.
You are a specialist in finding Indian government and official policy documents. You navigate Indian government web infrastructure, parliamentary records, and regulatory portals to locate primary sources for policy research.
You are a grant writing specialist for think tank and policy research organizations. Your role is to produce compelling, well-structured grant proposals that are grounded in evidence, aligned with funder priorities, and built on a defensible theory of change.
You are a literature review specialist operating inside an Obsidian project knowledge base.
You are a literature review specialist focusing on public policy research, with deep knowledge of Indian policy context. Your primary role is to conduct systematic literature reviews across academic papers, government reports, and think tank publications; identify research gaps; and help researchers formulate research questions and plans. You leverage Zotero as the central reference management system for paper collection, organization, full-text analysis, and citation export.
Use when creating or configuring Claude Code agents and their frontmatter.
Use this skill when the user asks to "critique this argument", "poke holes in this", "play devil's advocate", "hostile review", or as part of pre-submission review. Adversarially reviews a draft — identifies logical vulnerabilities, factual gaps, unstated assumptions, and missing counterarguments. Every critique comes with a path to fix it. Use mode:expert for experienced researchers (skips taxonomy explanations, goes straight to FLAW/FIX).
Use this skill when the user asks to "analyse this policy", "apply Bardach", "8-step policy analysis", "eightfold path", or "systematically analyse a policy problem". Walks through Eugene Bardach's 8-step methodology with Takshashila's India-specific lenses layered in at each step. Guided mode by default (asks questions, walks through all 8 steps); modular on request (jump to a specific step). Produces a Policy Analysis Memo at the end.
Use this skill when the user says "analyse the budget", "budget analysis", "how much did India spend on X", "compare BE vs actuals", "state budget for Y", "track scheme spending", "budget data for Z ministry", or wants to extract analytical insight from Union or State government budget documents. Guides the analyst from naming a budget to document extraction, ratio computation, chart specifications, and story angles. Does not write the final piece — hands off to op-ed-writing, policy-brief-writing, or discussion-document-writing.
Use this skill when the user asks to "map the causal logic", "show the causal loops", "make my argument's theory of change explicit", "causal analysis", or as part of pre-submission review. Extracts implicit causal claims from a policy argument, renders them as a Mermaid diagram with named loops, cross-connections between loops, layer models, and structural actor positions. Flags unsupported links and surfaces unintended consequences. Matches Takshashila publication standard.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
Has parse errors
Some configuration could not be fully parsed
Has parse errors
Some configuration could not be fully parsed
Structured policy reasoning tools for Takshashila researchers. Works as a Claude Code plugin. Embeds Bardach's 8-step analysis, causal loop diagrams, stakeholder mapping, and draft review — with India's institutional context built in at every step.
→ Student or outside user? See README-STUDENT.md
The three highest-value entry points for an experienced researcher:
1. Stress-test a draft before it goes to an editor
/draft-review
Paste your draft. Runs argument critique (expert mode), values review, and causal analysis. The argument critique will find the 2–3 things your editor or a hostile peer reviewer would raise — before they do.
2. Map the causal claims in a piece you are writing
causal analysis of [paste your argument]
Produces a Mermaid diagram with named feedback loops (R1 "ecosystem flywheel" etc.), cross-connections between loops, and a ranked list of which causal links are weakest. Good for discussion documents and policy briefs before submission.
3. Find committee reports you may have missed
/parliament-search [topic]
Searches all 24 DRSCs (16 Lok Sabha + 8 Rajya Sabha). Surfaces the most relevant reports immediately — not just a list, but "report #3 is most relevant because it contains data on X." Requires parliamentwatch installed locally.
Path A — You have a hypothesis (something you want to argue):
[H] hypothesis-development → Research Brief: falsifiable claim,
causal skeleton, assumption map,
evidence requirements
↓
[2] SOURCES /parliament-search + government-source-finder + /zotero-review
↓
[3] ACTORS stakeholder-analysis → Interest × power matrix
↓
[4] CAUSAL MAP causal-loop-analysis → named loops, cross-connections, leverage points
↓
[5] DRAFT /op-ed /policy-brief /grant-proposal discussion-document-writing
↓
[6] REVIEW /draft-review ← available at any stage
↓
[7] DISSEMINATE /promote
Path B — You have a policy problem (still working out what you think):
[P] /policy-analysis → Policy Analysis Memo: alternatives,
criteria, trade-offs, recommendation
↓
[H] hypothesis-development ← CONVERGENCE POINT
Crystallise the recommendation into a testable claim before drafting
↓
[3] → [4] → [5] → [6] → [7] same as Path A
The key question: "Can you state your argument in one sentence?"
hypothesis-development)/policy-analysis)/scholar asks this and routes you. /draft-review works at any stage.
A browser version of this toolkit is available at takshashilainst.github.io/scholar-web — no terminal, no install, just an API key.
It covers the core research lifecycle (13 of the workflows above) plus one tool that is web-only for now:
It covers the core research lifecycle (13 of the workflows above). Budget Analysis is available in both — the CLI version adds parliament search for CAG/PAC reports, direct URL fetching via defuddle, and Zotero integration.
The CLI plugin has deeper capabilities (parliament search, Zotero/Obsidian integration, voice calibration, grant scaffolding). The web app is for researchers and students who do not work in a terminal.
Takshashila Scholar is built on the four commitments of the Takshashila Institution:
The takshashila-values-review skill checks any draft against all four. The rules/takshashila-language.md file enforces vocabulary standards: Indian subcontinent not South Asia, West Asia not Middle East, no jargon.
Claude Code is Anthropic's AI assistant for your terminal — think of it as Claude with access to your files and persistent instructions. Install it once:
npm install -g @anthropic/claude-code
You need an Anthropic account. The free tier works for most research tasks.
Install the plugin:
npx claudepluginhub pranaykotas/takshashila-scholar --plugin takshashila-scholarProduction-grade academic research pipeline for Claude Code: research → write → review → revise → finalize. 4 skills, 27 modes, 39-agent ensemble, v3.7.3 + v3.8 L3 claim-faithfulness gate, v3.9.0 cross-index triangulation, v3.10 triangulation policy layer, v3.11 deterministic citation verification gate (#182).
PhD-level research capabilities: literature review, multi-source investigation, critical analysis, hypothesis-driven exploration, quantitative/qualitative methods
Multi-agent orchestrator for academic writing: 12 specialist agents and 30 writing principles for review, research, drafting, polishing, bibliography auditing, and literature surveys.
Specialized research analysis agents for critical thinking, evidence verification, synthesis, and parallel paper analysis
Academic research agents — hypothesis generation, experiment design, paper drafting, peer review simulation, and more.
Comprehensive Research Planning agents specializing in synthesising hypothesis and claims, researching related work and challenging assumptions.