Open-Source AI Skill for Claude

Turn Claude into an Evaluation Methodologist

PRAXIS is an open-source skill that gives Claude deep expertise in programme evaluation — from designing rigorous impact evaluations to building validated data collection instruments, adapted for real-world complexity.

claude — praxis-skill
// Ask Claude with PRAXIS loaded
> "Design a quasi-experimental
  evaluation for a PVE programme
  in the Sahel with limited baseline"
 
PRAXIS evaluation-designs.md
Loading DiD, PSM, ITS, realist methods...

Evaluation is hard. AI alone isn't enough.

General-purpose AI doesn't understand evaluation methodology. PRAXIS bridges that gap.

⚠️

Without PRAXIS

  • Generic AI gives surface-level evaluation advice that wouldn't survive a peer review
  • No awareness of OECD-DAC criteria, validated scales, or FCV-adapted methods
  • Cannot distinguish between a DiD and a regression discontinuity — let alone recommend one
  • Produces cookie-cutter surveys with no psychometric grounding

With PRAXIS

  • Claude becomes a methodologist who asks the right questions before recommending designs
  • 20+ evaluation approaches — experimental, quasi-experimental, theory-based, participatory, realist
  • 16+ validated scales (FCS, PHQ-9, WHODAS, WEAI) embedded and ready to deploy
  • Methods adapted for fragile, conflict-affected, and violent settings

From Burundi to the Sahel and Sri Lanka

Built on 12+ years of field experience across six countries, from rural Burundi to the Sahel.

📐

Evaluation Design

Experimental, quasi-experimental, theory-based, participatory, and complexity-responsive approaches. PRAXIS recommends designs based on your context, constraints, and evaluation questions.

20+ Approaches
📋

Data Collection Instruments

Surveys, FGD and KII guides, observation checklists — all grounded in psychometric principles. Includes 16+ validated scales ready for adaptation.

16+ Validated Scales
🔬

Realist & Case-Based Methods

CMO configurations, process tracing, QCA, contribution analysis — designed for evaluating complex programmes where RCTs aren't feasible or appropriate.

Complexity-Responsive
🔥

FCV-Adapted

Methods specifically adapted for fragile, conflict-affected, and violent settings. Accounts for access constraints, security risks, and rapidly shifting contexts.

Field-Tested
📊

Analysis & Reporting

Analysis plans, qualitative coding frameworks, synthesis methods, and quality assurance. Outputs ready for donor review and stakeholder communication.

End-to-End
🎓

Adaptive Expertise

Adjusts depth based on your experience level — from plain-language guidance for junior M&E officers to advanced causal inference for senior evaluators.

3 Levels

Up and running in three steps

PRAXIS is a Claude skill — a structured knowledge base that gives Claude domain expertise without fine-tuning.

Download the Skill

Clone or download the PRAXIS repository from GitHub. It's a collection of Markdown reference files — no dependencies, no API keys, no build step.

Add to a Claude Project

Upload the PRAXIS files to a Claude Project's knowledge base. Claude will automatically reference them when you ask evaluation questions.

Start Evaluating

Ask Claude to design an evaluation, build instruments, review a ToR, or recommend methods. PRAXIS handles the rest — with the rigour of a senior evaluator.

Who uses PRAXIS

Whether you're commissioning your first evaluation or designing your hundredth, PRAXIS meets you where you are.

🏠

M&E Teams

Design evaluations, build instruments, and produce donor-ready outputs — without hiring an external consultant for every task.

🎓

Researchers & Evaluators

Rapid methodological reference. Quickly compare designs, generate evaluation matrices, and stress-test your approach.

🌎

Organisations, Banks & Foundations

Strengthen evaluation quality across your portfolio. Give every programme team — from development banks to private foundations — methodological expertise on demand.

Try the Evaluation Approach Finder

Explore which evaluation design fits your context — powered by PRAXIS logic.

Evaluation Approach Finder
Click to launch the interactive tool

Better Evaluations,
Better Evidence

PRAXIS is free, open-source, and built for the evaluation community. Download it today and give Claude the methodological depth your programmes deserve.

Get PRAXIS on GitHub → Download v1.0.0