AI Analyst plugin
A set of commands, skills and subagents for Claude Code that runs data analysis like a careful team analyst.
- 326 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The agent reads your data and runs analysis code on your machine.
Install
- Works in
Manual install
git clone https://github.com/ai-analyst-lab/ai-analyst.gitThe Ten-minute start section: clone, run `pip install -e ".[dev]"`, start `claude` and enter `/setup`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent reads your data and runs analysis code on your machine.
This is third-party code. Review the repository files before installing.
What it does
AI Analyst turns Claude Code into a data analyst. It first asks which decision the question serves, profiles data before trusting it, pairs every number with a comparison, traces findings to rows and validates them before presenting. The repository holds 64 skills, 40 agents and 151 helper modules on Python 3.10 or newer. Work starts with `/setup` inside `claude`. There are public examples, synthetic NovaMart data and evaluation tasks; no private company data is included.
Who it is for. Product and business analysts who want verified conclusions from Claude Code on their own data rather than quick charts.
Good fit when
- You need to explain a conversion drop with period comparisons.
- Data must be profiled and gaps found before a report.
- A finding must be traced back to source rows.
- Analyst corrections should persist across sessions.
Not a fit when
- You need native work in Cursor or Codex.
- Data cannot go to a third-party agent.
- You only need one quick chart with no checks.
Example request
/setup, then: why did week 42 give fewer orders than week 41? Show the comparison and the rows it rests on.Limitations
It runs inside Claude Code and needs Python 3.10 or newer and a package install with `pip install -e ".[dev]"`. The set is large: 64 skills and 40 agents load context and spend tokens, so start with `/setup` and one question. Queries and data go to Claude; consider company policy. The last commit was in September 2026. A Cowork version lives in a separate plugin repository.
How to disable. Delete the clone and uninstall the package with `pip uninstall`. Correction memory lives in project files; delete it separately.
FAQ
Where do I start?
Clone, install the package, start `claude` and run `/setup`.
Is company data included?
No. Synthetic NovaMart data and public examples only.
How many skills are in the set?
Per the README 64 skills, 40 agents and 151 helper modules.
Is there a Cowork version?
Yes, in the separate ai-analyst-plugin repository.
How does the analyst check itself?
It profiles data first, compares every number and validates results before presenting.
Discussion
No questions or reviews yet.
Related
PluginsA team of data analysis agents that takes a dataset from raw rows to a trained model and checks its own conclusions.
31 skills for Claude Code and Codex for the analyst workflow: data checks, cohorts, funnels, A/B tests, dashboards, summaries
StatsClaw assembles nine Claude Code agents to build statistical packages from a paper or formulas with isolated pipelines.
26 skills and four reviewer agents for science in Claude Code: figure and statistics review, literature search, work handoff.
A Claude Code agent for protein design: target research, antibody and nanobody generation, screening and candidate ranking.
A template for data scientists: AGENTS.md, CLAUDE.md and skills for Python, SQL, Polars and model diagnostics.
README in short
The repository defines an analytical process for Claude Code: decision framing, profiling, comparison, row tracing and validation.