Stats Explorer MCP
Stats Explorer MCP lets an AI assistant run verified statistics, analytics, A/B tests, charts, and reports on local tabular data without pasting data into chat.
Inspect CSV/TSV/Excel/Parquet files and choose appropriate statistical tests.
Run guided hypothesis tests in six ordered steps: state hypotheses, check assumptions, check spread, power analysis, run test, and base-rate check, with skip reasons where allowed.
Perform t/z tests, ANOVA with Tukey HSD, ANCOVA, MANOVA, chi-square, Fisher exact, binomial, correlation, and A/B tests.
Analyze A/B/n experiments: conversion or continuous metrics, lift intervals, sample-ratio mismatch checks, Holm-adjusted comparisons, Bayesian probabilities, segment breakdowns, and sample-size planning.
Compute descriptive stats, confidence intervals, correlations without p-values, grouped summaries/top N, and "what drives X" relationship rankings.
Create charts (histogram, box plot, scatter, bar, line, pie) and build HTML/Markdown reports.
Handle very large files with streaming, full-read or sample choice, background jobs, progress, cancellation, and optional column caching.
Require user confirmation for columns/groups/charts/reports and optionally integrate Jev for independent checks.
Integrates with GitHub Copilot in VS Code so Copilot Chat can call Stats Explorer's tools for statistical tests, A/B analysis, exploration, charts, and reports on local data files.
Supports production orchestrated workflows where LangGraph signs Stats Explorer decisions via an approval secret, enabling auditable approval of test plans and conclusions.
Supports orchestrated deployments where Temporal signs Stats Explorer decisions via an approval secret for production pipelines with audit logs.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Stats Explorer MCP Run an A/B test on conversion by variant in campaign.csv"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Stats Explorer MCP

An MCP server with 18 statistics and analytics tools an AI assistant can call instead of guessing numbers, on data files of any size:
Tests: t-tests, z-tests, ANOVA (with a Tukey HSD post-hoc that names the groups that differ), ANCOVA, MANOVA, chi-square, Fisher's exact, binomial and correlation.
A/B testing: A/B and A/B/n readouts for conversion rates and revenue-type metrics (lift with intervals, sample ratio mismatch check, Holm-adjusted comparisons, Bayesian probability to beat the control, segment breakdowns), and sample-size planning.
Analytics: totals, averages and counts by category (top N), and "what drives X?" rankings across every column.
Helpers: descriptive stats, confidence intervals, power analysis, a "which test should I use?" tool, and a data-file inspector.
The tools can read a developer's CSV, TSV, Excel or Parquet files by column name, so large datasets never get pasted into the chat. Each result includes:
the effect size and confidence interval,
a plain-English interpretation,
warnings when an assumption looks shaky.
On top of that:
Correct. 43 reference cases are checked against SciPy and statsmodels on every commit, to at least 6 significant figures.
Guided by default. The agent first decides whether your request needs a hypothesis test or just exploration. Hypothesis tests follow the six steps of Stats Explorer's Bring it all together guide, in order.
You stay in control. Every column, group, chart type and report section the agent proposes is shown to you to confirm or change before anything runs.
Private. Runs locally over stdio. It makes no network calls and needs no API key (unless you turn on Jev's enforce mode, below). Data files are read only from your workspace folders.
Checked by Jev, if you want. Connect Jev, TypeSafe's fast judgment model, alongside it, and Jev independently checks the test plan and the assistant's written conclusions against the computed results. You can run it without an orchestrator, enforced by Stats Explorer itself, or under a production orchestrator that signs every decision.
The engine, the file import and the six-step guide are shared with the Stats Explorer app, so results match the app exactly.
Connect it (pick your assistant)
GitHub Copilot in VS Code: about 1 minute, nothing else to install
In VS Code, open Extensions (Ctrl+Shift+X / Cmd+Shift+X), search Stats Explorer MCP, click Install.
Open Copilot Chat and switch the mode dropdown to Agent.
Click the tools icon in the chat box and check that Stats Explorer is listed and ticked.
Open the folder that holds your data and ask a question (see Example questions).
The extension registers the server itself, using VS Code's built-in runtime, so you don't need Node.js or an mcp.json file. Confirmations appear as forms you can edit.
Claude Code (terminal or the Claude Code panel in VS Code): about 2 minutes
You need Node.js 22 or later (24 LTS recommended). In a terminal, run this once:
claude mcp add -s user stats-explorer -e STATS_EXPLORER_CONFIRM_VIA=chat -- npx -y stats-explorer-mcp
claude mcp listclaude mcp list should show stats-explorer … ✓ Connected. Then start a new Claude session, because servers are loaded when a session starts. In VS Code: Ctrl+Shift+P → Developer: Reload Window, then New session. Type /mcp in the chat to see the servers the session has.
-s usermakes it available in every folder (without it, only in the folder where you ran the command).STATS_EXPLORER_CONFIRM_VIA=chatmakes confirmations appear as messages in the chat, which you answer by replying. Claude Code doesn't always display form-style prompts.If you have the VS Code extension installed, Stats Explorer: Copy Setup Command for Claude Code copies this command for you.
Claude Desktop, Cursor, Windsurf, Cline and other MCP clients
Add this to the client's MCP config (claude_desktop_config.json, .cursor/mcp.json, and so on), then restart the client:
{
"mcpServers": {
"stats-explorer": {
"command": "npx",
"args": ["-y", "stats-explorer-mcp"],
"env": { "STATS_EXPLORER_ROOTS": "/path/to/your/data", "STATS_EXPLORER_CONFIRM_VIA": "chat" }
}
}
}Requires Node.js 22 or later. STATS_EXPLORER_ROOTS is the folder (or folders) the tools may read; see Data files.
Example questions
Name the file in your message (or attach it); the assistant passes it to the tools and never pastes the data into the chat.
What's in sales.csv?
Total revenue by state in sales.csv, top 10. (group_summary)
What drives profit in sales.csv? (find_relationships, then follow-ups)
Show the distribution of order value and a box plot by segment.
Does average profit differ between the Consumer and Business segments? Is it significant? (six-step hypothesis test)
Did the new checkout page (variant B) increase conversion in experiment.csv? Break it down by device. (ab_test)
How many visitors do I need to detect a 10% lift on a 4% conversion rate, with 3,000 visitors a day? (ab_test_plan)
Does revenue differ between the three regions, and which ones? (ANOVA + Tukey HSD)
Make an HTML report of what we found and save it as findings.html.
The analyze, explore_data and hypothesis_test prompts (slash commands in VS Code and Claude) start the same flows.
Related MCP server: mix_server
Working with Jev (optional)
When both Stats Explorer and a Jev MCP server are connected to Claude or Copilot, they split the work. Stats Explorer computes every number, and Jev checks two things before you see a result:
The test plan: is this the right test, are the hypotheses right, are these the right columns?
The written conclusions: does every number match, is "significant" stated correctly, are there claims of cause the analysis can't support?
Stats Explorer also runs its own number and significance checks on every claim. Only claims that pass go into reports, and each report lists every check under Independent checks.
Mode | Set | Who calls Jev | Good for |
Cooperate |
| The assistant, through your Jev MCP server, using packets Stats Explorer prepares | Working in Claude Code or Copilot without any orchestration tool |
Enforce |
| Stats Explorer, directly from the Jev API | Checks the assistant can't skip or alter |
Orchestrated |
| Your orchestrator (LangGraph, Temporal, Step Functions and so on), which signs each decision | Production pipelines with audit logs |
Quick start with Claude Code (no orchestrator):
claude mcp add -s user jev -e TYPESAFE_API_KEY=YOUR_KEY -- npx -y @jkudish/jev-mcp
claude mcp add -s user stats-explorer -e STATS_EXPLORER_JEV=cooperate -e STATS_EXPLORER_CONFIRM_VIA=chat -- npx -y stats-explorer-mcp
npx -y stats-explorer-mcp init-jev # adds routing rules to CLAUDE.md and .github/copilot-instructions.mdAny Jev MCP server works (@jkudish/jev-mcp, jev-mcp, typesafe-mcp, …): the packets come in the shapes their tools take. In VS Code, set Stats Explorer › Jev: Mode. See docs/JEV.md for every mode and setting, the signed-approval format (with JavaScript and Python), what is sent to Jev, and the runnable orchestrator example in examples/orchestrator/.
How it works (default)
1. The agent classifies your request
Every analysis starts with start_analysis, where the agent picks a mode:
Your request sounds like… | Mode | What happens |
"Does the drug change scores?", "Is A better than B?", "Is this difference significant?" |
| The six-step workflow below |
"Summarise…", "Show me the distribution of…", "Chart X by Y", "What's the average, with a CI?" |
| Descriptive tools and charts only: no hypotheses, no p-values, and no hypothesis section in the report |
If it's unclear, the agent is instructed to ask you. Nothing else runs until start_analysis has been called.
2. Hypothesis tests follow the six steps, in order
Step (from the app's Learn tab) | Tool |
1. State your question and choose the right test |
|
2. Validate your data against the test's assumptions |
|
3. Check spread |
|
4. Calculate power before you run |
|
5. Run the test, then report honestly |
|
6. Sanity-check rare events against base rates |
|
Calling a step out of order is refused with a message that names the next step. Only steps 4 and 6 can be skipped, and only with a reason that appears in the report. If step 2 shows the chosen test doesn't fit, re-stating the plan in step 1 is allowed and restarts the later steps. The app's guide calls this a principled revision, not p-hacking.

3. You confirm every selection
When the agent proposes columns, groups, a chart type or report sections:
Clients that support MCP forms (elicitation), such as VS Code, show you a form with the agent's choices pre-filled as dropdowns and checkboxes. Change anything, then accept or decline. Declining runs nothing.
Other clients get a
CONFIRMATION NEEDEDresponse listing the proposal, the alternatives and a one-time token. The agent is instructed to show you the choices and call again with the token only after you agree. A token only works for the exact selection it was issued for, so a changed selection is proposed again.
A selection you've confirmed isn't asked about twice in the same session.
4. Charts and reports
create_visualization: histogram, box plot, scatter (with least-squares line), bar (mean or count by category), line, and pie. These are the app's chart types. Returns an accessible SVG (light and dark mode) and can save it as.svg.build_report: only when you ask for one, and containing only the sections you tick. A completed hypothesis test becomes a write-up in the app's format: the question, the hypotheses, how it was tested (with steps 2–4), what was found, the verdict and base rates. Explorations never get a hypothesis section. Saved as HTML (default) or Markdown in your workspace; existing files are not overwritten unless you say so.


Images show real output on a sample dataset.
Tools
Tool | Purpose |
| Classify the request (hypothesis test or exploration) |
| Columns, types, missing values, categories and a preview of a data file |
| Recommend the right test for a question |
| The six steps |
| n, mean, SD, median, quartiles, range, skewness, kurtosis, shape, mean CI (optionally by group) |
| CI for a mean (t) or a proportion (Wilson) |
| Pearson or Spearman r with CI and fitted line, without a p-value (a significance test goes through step 1) |
| Totals, averages, medians, counts per category; top N (analytics) |
| Visitors and days needed for an A/B test, or the smallest lift a given sample can detect |
| Which columns relate most to a numeric outcome ("what drives X") |
| Charts and reports |
| Where the current analysis is, and the saved results |
| Very large files: read all or a sample, background jobs, kept columns |
Tests available in step 1: t-test (one-sample, paired, Welch or Student), z-test (mean with known σ, proportion), one-way ANOVA (with a Tukey HSD post-hoc, plus optional Bonferroni Welch pairs), ANCOVA, MANOVA, chi-square (goodness of fit, independence), Fisher's exact, exact binomial, correlation and A/B test.
ANOVA post-hoc (Tukey HSD). Every one-way ANOVA with 3 or more groups includes Tukey's HSD (Tukey–Kramer for unequal group sizes), on in-memory and very large files alike: the difference, adjusted p-value and simultaneous interval for every pair, the significant pairs marked ★ ("B > A"), the best group and whether it beats all others, and a letter display (groups sharing a letter don't differ). Reports name the groups that differ.
A/B tests (ab_test). From inline numbers or a file with a variant column and an outcome column:
metric: "conversion"(converted yes/no): two-proportion z-test, with Fisher's exact p when counts are small.metric: "continuous"(revenue, time, order value): Welch's t-test.Every variant against the control: absolute lift (percentage points) and relative lift, each with an interval; Cohen's h or d.
Sample ratio mismatch check against the planned split (
expected_split): a broken assignment blocks the decision.A/B/n: an omnibus test (chi-square, or ANOVA with Tukey HSD) and Holm-adjusted p-values (or Bonferroni / none).
Bayesian view: probability each variant beats the control, probability of being best and the expected loss of shipping it (Beta-Binomial, computed by quadrature, so the numbers are reproducible).
The minimum detectable effect of the test as run, an optional
mde(smallest lift that matters) to judge practical significance, an optionalsegment_columnbreakdown with a Simpson's-paradox warning, and a plain decision.In the guided workflow, step 2 adds the sample ratio check and step 4 sizes power in lift terms.
The server also offers the prompts analyze, hypothesis_test and explore_data, which clients such as VS Code show as slash commands.
Data files
Stats Explorer is built for tabular formats (CSV, TSV, Excel and Parquet) to run hypothesis tests or explore statistics. Files are read with the app's own import code:
quoted fields, embedded commas and line breaks, BOMs, and tab, semicolon or pipe separators;
missing values (blank, NA, N/A, null, #N/A, …) are skipped;
numbers are parsed strictly (
"12 kg"is text, not 12), while1,234,$12.50and45%still count as numbers;dates become
YYYY-MM-DDtext;duplicate column names are renamed rather than overwritten;
Excel serial dates are converted;
Parquet files compressed with Snappy, gzip, zstd, Brotli or LZ4 are all read.
There is no size limit for CSV, TSV and Parquet files; see Large files.
JSON, BSON, NDJSON and NoSQL databases (MongoDB, CouchDB, Redis connection strings, …) are document data, not tables. For these, every tool answers with that statement and suggests exporting to CSV or Parquet first (for example mongoexport --type=csv). Inline rows containing nested objects or arrays get the same answer.
Files are read only from allowed folders, and symlinks that point elsewhere are refused:
STATS_EXPLORER_ROOTS, if set (separated by:on macOS/Linux,;on Windows), otherwisethe workspace roots your MCP client reports, otherwise
the server's working directory.
Large files (gigabytes to terabytes)
Files up to 512 MB are loaded into compact column storage (a 190 MB, 1,000,000-row CSV loads in about 8 seconds using under 0.5 GB of memory). Larger CSV, TSV and Parquet files have no size limit. They are streamed instead, in a single pass that every tool needs:
One pass, fixed memory. Each tool reads only the columns it uses, in parallel pieces (one per CPU core), and keeps running totals: counts, means, variances, sums of squares and cross-products. Memory does not grow with the file: about 0.5 GB with one reader, plus about 0.2 GB per extra parallel reader. A 2.5 GB, 13-million-row CSV took 45 seconds on 1 core and 22 seconds on 2; the same data as a 340 MB Parquet file took 9 seconds.
Exact where it can be. t-tests, z-tests, ANOVA (with pairwise comparisons), ANCOVA, MANOVA, chi-square, Fisher's exact, Pearson correlation, means, standard deviations, skewness, kurtosis and confidence intervals are exact on every row. Groups of up to 100,000 values keep their actual values, so everything about them is exact, medians included.
Estimated where it must be, and labelled. Medians, quartiles, histogram bars and outlier counts of larger groups come from a KLL quantile sketch (rank error about 0.03% in testing). Spearman's ρ comes from a random sample of 100,000 pairs, and
inspect_data_file's count of different values in a column, above 10,000, from HyperLogLog (within about 2%). Results that contain estimates list them inlarge_file.approximateand in their warnings.Asked first. If reading the whole file would take more than 2 minutes, you are asked whether to read every row (exact) or a quick sample of about 1% of the rows in blocks spread across the file (fast, approximate). Clients with forms show the choice as a form; otherwise the assistant asks you and calls
large_file_plan.Runs in the background. A read that takes longer than 40 seconds continues as a background job; the tool answers
STILL RUNNINGwith ajob_id, andjob_statusreturns the result when it's ready (with progress and time left meanwhile).cancel_jobstops it. Clients that support MCP progress notifications see a progress bar.Read once. The workflow steps share one pass: step 1 reads the file, and steps 2 to 5 reuse the totals. With your consent, the columns read can be kept on disk (8 bytes per number, 4 per text value), so later questions on them take seconds.
manage_cachelists or deletes them.Same rules as small files. The streaming reader applies the app's import rules exactly. It was checked against the in-memory reader on 24 edge-case files and 300 random ones, and on 228 tool calls (tests, charts and full six-step workflows) across CSV and Parquet files, split into as many as 3 parallel pieces. Line breaks inside quoted values are handled: if a split point falls inside one, the file is re-read in one piece.
Excel files are always read whole (up to about 1.9 GB); export very large workbooks to CSV or Parquet.
Configuration
Variable | Default | Effect |
|
|
|
|
|
|
| workspace | Folders data files may be read from (and reports written to) |
|
| CSV/TSV/Parquet files above this size are streamed (Parquet: by rows × columns) |
| cores − 1 | Parallel readers for large files (at most 16 by default) |
|
| Ask before a full read expected to take longer than this |
|
| Continue as a background job after this long |
|
| Size of the quick sample |
|
| Keep columns read from large files: |
| system temp | Where kept columns go |
| no limit | Refuse files larger than this |
|
|
|
|
|
|
| Changes that link | |
|
| Jev checks: |
Accuracy
test/reference/generate.py computes expected values with SciPy and statsmodels for 43 cases. They cover every test and alternative, Welch and Student t-tests, p-values below 1e-30, ties in Spearman, ANCOVA (Type II SS), MANOVA with 2 and 3 outcomes, and file inputs with missing values. test/reference/guided.json adds the ANCOVA slope-equality F test and the step-4 power calculations (two-sample, one-sample, proportion and correlation), checked against statsmodels. test/reference/abtest.py checks Tukey HSD (studentized range, adjusted p-values, intervals and rejections against scipy.stats.tukey_hsd and statsmodels' pairwise_tukeyhsd) and every A/B number: z, p and one-sided p (proportions_ztest), lift intervals (confint_proportions_2indep), Wilson intervals, Fisher's exact, chi-square, Holm/Bonferroni (multipletests), the sample ratio check, Welch t-tests, the Bayesian probabilities and expected loss (numerical integration in SciPy) and sample sizes (NormalIndPower). npm test runs everything through real MCP clients over stdio (with and without form support) and requires agreement to a relative error of 1e-6.
Known approximations, also stated in the results:
power uses the normal approximation (within about 0.01 of exact power, matching statsmodels'
NormalIndPower);Spearman p-values use the t approximation (as SciPy does);
chi-square tests use no continuity correction;
on very large files, medians, quartiles, histogram bars, outlier counts and Spearman's ρ are estimated (see Large files).
Limits worth knowing
Confirmation forms appear only in clients that support MCP elicitation. Elsewhere the token flow relies on the agent relaying the proposal to you, as the server instructs. A token can't be guessed, and it can't be reused for a different selection.
Which mode a request needs is decided by the agent, guided by the server's instructions. If it picks the wrong one, say so; it can start a new analysis.
Development
npm install
npm test # reference, unit, guided-workflow and large-file tests
STATS_EXPLORER_STREAM_MB=0 npm test # the same, every file through the streaming path
STATS_EXPLORER_STREAM_MB=0 STATS_EXPLORER_EXACT_CAP=3 npm test # ... using only one-pass summaries
npm run inspect # open the MCP Inspector
npm run reference # regenerate reference cases (needs Python + scipy + statsmodels)
npm run reference:ab # regenerate the Tukey / A/B reference values and test/fixtures/experiment.csv
npm run jev:eval -- --dry-run # measure Jev on labeled cases (drop --dry-run and set TYPESAFE_API_KEY to call Jev)
cd vscode-extension && npm install && npm test && npm run package # build the .vsixsrc/engine/ is copied from the Stats Explorer app by npm run sync-engine -- <path-to-app>: the statistics, the data import module and the Learn tab's workflow text. Fix them in the app and re-sync, so the two never drift.
License
MIT
Available Tools
23 toolsab_test_planA/B test sample sizeARead-onlyIdempotent
Plan an A/B test: visitors needed per variant to detect the smallest lift that matters (mde) at a given power, and/or the smallest lift detectable with a given n_per_variant; days needed from daily_traffic. Conversion: baseline_rate. Continuous: baseline_sd (and baseline_mean for a relative mde). Use in an exploration (planning a future test); a running hypothesis test sizes itself in step 4.
Inputs: metric (required), mde, alpha, power, +9 more.
base_rate_checkStep 6: sanity-check rare events against base ratesARead-only
Step 6 (or standalone in an exploration). For rare-event questions (fraud, defects, diagnoses, churn): what share of positive flags is real (PPV). Give prevalence, sensitivity, specificity (0–1), or file + condition_column + test_column (0/1 columns). If the question is not about rare events, use skip_step 6 instead.
Inputs: file, prevalence, sensitivity, specificity, +3 more.
build_reportBuild a reportA
Builds a report from results created in this session, only when the user asks for one. include = the result ids the user wants (default: the current analysis). The user confirms the sections. A hypothesis-test write-up (question, hypotheses, how it was tested, findings, verdict, base rates) appears only for completed hypothesis tests; explorations get no hypothesis section. format html (default) or markdown; save_to writes a file in the workspace.
Inputs: title, format, include, save_to, +2 more.
cancel_jobCancel a large-file jobBRead-onlyIdempotent
Stops a background job started by a tool that answered STILL RUNNING.
Inputs: job_id (required).
check_assumptionsStep 2: validate the data against the test's assumptionsARead-only
Step 2. Checks shape (skew, kurtosis, small samples), group sizes, outliers, expected counts, ANCOVA slope equality and missing values for the test fixed in step 1. If an assumption fails, revising the test now (state_hypotheses again) is principled.
Inputs: no named parameters.
check_spreadStep 3: check spreadARead-only
Step 3. Standard deviation, IQR, coefficient of variation, outliers and variance ratio for the data in the plan.
Inputs: no named parameters.
choose_testChoose a statistical testARead-onlyIdempotent
Recommends which test tool to run for a research question, with the assumptions to check. Call this first when unsure.
Inputs: goal (required), paired, has_covariate, multiple_outcomes, +3 more.
confidence_intervalConfidence intervalCRead-only
CI for a mean (column the user asked about) or a proportion. The user confirms the column first.
Inputs: type (required), n, file, column, +4 more.
correlationCorrelation (descriptive)ARead-only
Strength and direction of the relationship between two numeric columns (r or ρ, CI, least-squares line), WITHOUT a p-value. To test whether a relationship is significant, use a hypothesis_test analysis with test "correlation". The user confirms the columns first.
Inputs: x, y, file, method, +6 more.
create_visualizationCreate a chartA
Creates a chart for the columns the user asked to see: histogram (column), box_plot (column, optional group_column), scatter (x_column, y_column), bar (x_column categories, optional y_column mean), line (x_column, y_column), pie (x_column categories, optional y_column sum). The user confirms the chart type and columns first. Optional save_to (.svg). Charts are kept for build_report.
Inputs: file (required), chart (required), bins, title, +7 more.
describeDescriptive statisticsBRead-only
n, mean, SD, median, quartiles, range, skewness, kurtosis, shape and CI for the mean, for a column the user asked about (optionally by group). The user confirms the columns first.
Inputs: file, name, column, values, +3 more.
find_relationshipsWhat relates to a columnARead-onlyIdempotent
For open analytics questions such as "what drives revenue?", "what is related to churn time?" or "which factors matter for score?": ranks every other column (or the ones given) by how strongly it relates to a numeric target column. Pearson r for numeric columns, correlation ratio η plus the highest/lowest group for categorical ones. Descriptive (no p-values); follow up on the top findings with group_summary, correlation, charts or a hypothesis test. Works on files of any size. Use it in an explore analysis. The user confirms the target and the columns first.
Inputs: file (required), target (required), top, columns, +1 more.
group_summarySummary by groupARead-only
Analytics-style breakdown: count, sum, mean, median, min, max or SD of a numeric column for each category of another column (e.g. total revenue by region, average order value by segment, top 10 products by profit, rows per status), sorted, top N. Without value_column it counts rows per category. Descriptive only, no p-values. The user confirms the columns first.
Inputs: group_column (required), top, file, stats, +4 more.
inspect_data_fileInspect a data fileARead-onlyIdempotent
Lists the columns of a local CSV/TSV/JSON file with their types, missing values, ranges and category values, plus a 5-row preview. Use before running tests on a file so you pass correct column and group names.
Inputs: file (required).
job_statusWait for a large-file jobARead-onlyIdempotent
For a tool that answered STILL RUNNING: waits up to wait_seconds for the background job, then returns its result (the same result the tool would have returned) or its progress so far.
Inputs: job_id (required), wait_seconds.
large_file_planChoose how to read a large fileARead-onlyIdempotent
Only when a tool answered LARGE FILE, and only after asking the user: record their choice for that file. choice "full" reads every row (exact results, can take a long time, runs in the background); "sample" reads about 1% of the rows in blocks spread across the file (quick, approximate). save_columns keeps the columns that are read on disk so later questions on them are fast. Then call the original tool again.
Inputs: file (required), choice (required), save_columns.
manage_cacheSaved large-file columnsBDestructiveIdempotent
Lists or deletes the columns kept on disk from large files (see large_file_plan save_columns). action "list" or "clear" (all, or one file).
Inputs: action (required), file.
power_analysisStep 4: calculate power / plan sample sizeARead-only
In a hypothesis test (step 4): power of the planned test for the smallest effect the user cares about, using the data's sample size. smallest_effect is Cohen's d for means, r for correlations, or the true proportion p₁ for proportion tests. Ask the user for it; if they don't know, pass use_conventional: true. In an exploration: give two of effect_size, n_per_group, power to plan a two-group study.
Inputs: alpha, power, alternative, effect_size, +3 more.
run_testStep 5: run the testARead-only
Step 5. Runs exactly the test, columns and hypotheses the user confirmed in step 1 and returns the statistic, p-value, effect size, CI and verdict. Requires steps 1–4.
Inputs: no named parameters.
skip_stepSkip step 4 or 6 with a reasonARead-only
Only steps 4 (power) and 6 (base rates) can be skipped, with a short reason that will appear in the report.
Inputs: step (required), reason (required).
start_analysisStart an analysisARead-only
Call this first for every new question. Read the user's request and choose mode: "hypothesis_test" when they want to know whether a difference, effect or relationship is real / significant / caused by something, or want to test a claim; "explore" when they want to describe, summarise, visualise, estimate or plan, without testing a claim. If the request is ambiguous, ask the user before calling. Returns the steps to follow.
Inputs: mode (required), question (required), file.
state_hypothesesStep 1: state the question, hypotheses and testARead-only
Step 1 of a hypothesis test. Fix the test, the data selection and H₀/H₁ BEFORE looking at results. test_arguments are the same fields the test takes (e.g. t_test: {type, file, column, group_column, group1, group2, equal_variances, mu0}; anova: {file, column, group_column}; ab_test: {metric, file, variant_column, outcome_column, control, success_value, expected_split, segment_column}; chi_square: {type, file, row_column, col_column}; correlation: {file, x_column, y_column, method}). Only use columns the user asked about. The user confirms every column/group choice, the direction and α. Calling it again later revises the plan and restarts steps 2–6.
Inputs: test (required), test_arguments (required), h0, h1, +4 more.
workflow_statusWorkflow statusARead-onlyIdempotent
Shows the current analysis, which of the six steps are done, and the saved results (for the report).
Inputs: no named parameters.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 23 tools
The six-step hypothesis workflow is explicitly numbered, and the tool descriptions draw deliberate lines (e.g. correlation vs. correlation hypothesis test, describe vs. group_summary). However, start_analysis and choose_test both claim to be "call this first," and describe/group_summary/confidence_interval overlap on summary statistics, so a few boundaries need careful reading.
All names are consistently snake_case and readable, but conventions mix verbs (run_test, start_analysis, find_relationships) with bare nouns (describe, correlation, workflow_status, skip_step). The pattern is predictable enough to navigate, just not uniform verb_noun.
23 tools is on the heavy side for a stats assistant; several workflow steps (check_assumptions/check_spread, skip_step, workflow_status) are granular enough that they could be folded into parameters or a single status tool. Each is individually justified by the guided methodology, so nothing is pure filler, but the surface is larger than necessary.
Coverage of the analysis lifecycle is strong: inspection, descriptives, assumption checks, power, tests, planning, visualization, reporting, plus caching and background-job management. Gaps exist around row filtering/derived columns and listing or exporting raw datasets, which limit deeper exploratory work but are workable around.
Maintenance
Related MCP Connectors
A statistical analyst in your AI chat: bring data and a question, get a citable, re-runnable report.
Transform your data analysis with our Data Compute & Stats Bot. Effortlessly calculate descriptive
Open, inspect, filter, edit and convert xlsx and csv files from your AI chat. Processing is local.
Open, inspect, filter, edit and convert xlsx and csv files from your AI chat. Processing is local.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables conversational analysis of CSV and Parquet files through natural language, providing statistics, summaries, data type information, and comprehensive multi-step data analysis.-
- FlicenseNot gradedqualityDmaintenanceEnables AI assistants to interact with local CSV and Parquet data files through natural language queries, facilitating tasks like summarizing datasets or retrieving specific information.4-
- AlicenseAqualityCmaintenanceEnables AI assistants to analyze CSV files by providing statistical analysis and structural summaries via MCP tools.2MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to perform data analysis on local datasets through natural language, orchestrating Python, R, SQL, Spark, Jupyter, Power BI, and Tableau engines for statistics, machine learning, dashboard generation, and batch processing.Apache 2.0