Analysis Kit
stephs-repos/analysis-kit
Scaffolding agentic para análise de dados com validação de claims integrada
Ver no GitHubFramework de scaffolding para projetos de análise de dados agentic com Claude Code. Garante integridade de claims quantitativos através de validação de saída, contrato de confiança e rastreabilidade completa de cada asserção.
Instalação
npx convez-ai stephs-repos/analysis-kit1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Projetos de análise de dados com rastreabilidade de claims
- Validação automática de asserções quantitativas antes de commits
- Documentação viva e reprodutibilidade de análises
- Auditoria de decisões analíticas com counterfactual tagging
Veredito

Framework maduro e bem documentado para quem precisa de rigor científico em análise de dados agentic. Opinião forte sobre disciplina (claims com IDs, code paths, caveats) — excelente para stakeholders que demandam auditabilidade; overkill para explorações rápidas.
README
A tradução para o português deste README ainda está na fila.
analysis-kit
A scaffolding framework for agentic data-analysis projects with Claude Code as the operator.
The framework's job is claim integrity: every quantitative assertion you make to a stakeholder is backed by a finding ID, a reproducing code path, declared filters, and a counterfactual tag. Validation runs as exit-code; hooks block bad commits; templates carry caveats forward.
This is scaffolding, not a runtime library. Once a project is created, it has no runtime dependency on analysis-kit — upgrades are deliberate, opt-in copies of the templates.
Documentation
- 🗺️ CONCEPTS.md — start here. The mental-model map: the main features in plain language and how they relate, with diagrams.
- 📖 USER_GUIDE.md — the detailed hands-on tour: how validation works, every template explained, common workflows, troubleshooting, when not to use.
- PHILOSOPHY.md — principles
- PROVENANCE_CONTRACT.md —
findings.jsonschema reference - COUNTERFACTUAL_TAGGING.md —
OBSERVED/PLAUSIBLE/WEAKrules - HOOKS_GUIDE.md — hook contracts and failure modes
Quick start
# Clone analysis-kit somewhere
git clone <your-fork-url> ~/dev/analysis-kit
# Create a new analysis project
~/dev/analysis-kit/bootstrap/new-project.sh ./my-analysis --minimum
cd my-analysis
pip install -r requirements.txt
claude # opens Claude Code; reads CLAUDE.md → follows the discipline
Claude Code skills (optional, recommended)
Install three workflow skills (/akit-start, /akit-fill, /akit-finding) plus an index (/akit) globally so Claude Code can guide users through the analysis-kit lifecycle:
~/dev/analysis-kit/bootstrap/install-skills.sh
Then in any new project: /akit-start <name> → drop reference materials → /akit-fill → /akit-finding "<hypothesis>" for each claim. See skills/ for the source markdown.
VS Code dev container (recommended for working on analysis-kit itself)
The repo ships a .devcontainer/ config so you can open it in VS Code with a reproducible environment in one click:
- Open the repo folder in VS Code.
- When prompted ("Reopen in Container"), accept — or run
Dev Containers: Reopen in Containerfrom the command palette. - Wait for the build (~2 min first time). Post-create runs
pytestas a smoke test.
You get Python 3.13, pinned pandas/pandera/numpy/pytest, Quarto, jq, and the gh CLI — all verified together. See .devcontainer/ for the exact pins.
Tiers:
--minimum— CLAUDE.md, six live-docs,validate.py,memory/,.claude/hooks/. Pandera in deps.--full— adds the Quarto vignette pipeline and_quarto.yml.
What you get
| Contract | File | Purpose |
|---|---|---|
| Trust gate | analysis/validate.py | Exit 0 = trustworthy. Stop hook fires --fast; commit hook blocks on full. |
| Claims ledger | analysis/output/findings.json | Every claim has F-NNN id, code_path, an input + reproducibility block, caveats, counterfactual_tag, measurement_ref. |
| Live docs | live-docs/*.md | TRUST_MEMO, DATA_PROFILE, DECISIONS, ANALYSIS_BACKLOG, TOOLING, METHODOLOGY_LOG — peers, all amendable. |
| Caveat carriers | memory/*.md | Preconditions the agent reads before aggregating. Templates declare shape, projects fill content. |
| Hooks | .claude/hooks/ | validate-on-stop (fast, blocks the turn on red with a loop guard), block-unvalidated-commit (full, fails closed), findings-coverage-on-edit (nudge). |
What this is not
- Not a pipeline runner. Use Kedro or Snakemake for orchestration.
- Not a notebook framework. Use Ploomber for notebook-first work.
- Not an MLflow replacement. Use MLflow for modelling.
- Not a BI tool. Use Evidence.dev for warehouse-resident BI.
It is dbt-tests-for-pandas-projects, with claims as the unit of trust, designed for an LLM operator.
Philosophy
In one sentence: constrain the agent's improvisational surface so deterministic guardrails can do the work. See docs/PHILOSOPHY.md for the full statement.
Status
v1.0.0. The contracts are stable. A finding declares its data dependency in an input block (sources with pinned content hashes, columns) and a reproducibility block (filters, post-filter row count); values can be registered by execution (register_computed) so a number can't drift from the code that produced it. The self-test suite passes in CI across Python 3.11–3.13. Validated end-to-end against a 35-finding production project (caught 3 stale findings the project's own validator hadn't flagged). Quarto integration shipped; Datasette deferred to a later release.