How do I parse and generate CSV data in an MCP pipeline without a runtime dependency on Python or external CLI tools?

@mukundakatta/csv-mcp v1.0.4 — CSV parse & stringify for MCP pipelines

Install: npm install @mukundakatta/csv-mcp Entry: node node_modules/@mukundakatta/csv-mcp/dist/server.js (stdio transport) Auth: none required

2 Tools

Key Findings

1. `header: true` returns objects, `header: false` returns arrays — with header: true, each row is {col: value} keyed by first-row headers; with header: false (default), rows are plain arrays of strings.
2. All values are strings — no auto-type coercion. Numbers, booleans, dates all come back as strings. Parse them yourself.
3. Custom delimiters work — pass delimiter: "\t" for TSV, delimiter: ";" for European-style CSV. Works in both parse and stringify.
4. Unicode-safe — Turkish characters (ğ, ü, ş, İ, ç), CJK, emoji all round-trip correctly.
5. RFC 4180 double-quote escaping in stringify output — fields containing commas, newlines, or quotes are wrapped in double quotes; embedded quotes become "". Example: she said "hi" → "she said ""hi""".
6. ⚠️ Backslash-escaped quotes in parse input get stripped — if you pass \" instead of RFC 4180 "", the backslash is silently dropped. The parser is not RFC 4180-strict on input; it's lenient but lossy for backslash escapes.
7. Round-trip verified — parse → stringify → parse produces identical data structures.
8. Empty input → `{rows: []}` from parse, `""` from stringify — no errors on empty data.
9. CRLF line endings in input are handled correctly — \r\n is treated the same as \n.
10. `eol` param in stringify — defaults to \n; pass eol: "\r\n" for Windows-style line endings.
11. Sub-millisecond latency — all 15 calls completed in 0ms (in-process, no network).

Verified Trace (15 calls, 100% success, p50=0ms)

• parse basic 3-col CSV → 2 rows of arrays ✓
• parse with header: true → objects keyed by header names ✓
• stringify objects with header: true → CSV with header row ✓
• parse with delimiter: "\t" → TSV parsed correctly ✓
• stringify with delimiter: "\t" → TSV output ✓
• parse semicolon-delimited CSV → correct split ✓
• parse Turkish Unicode text → ğ, ü, ş, İ, ç preserved ✓
• stringify Unicode rows → round-trips cleanly ✓
• parse quoted fields with embedded commas and newlines → correct ✓
• stringify fields requiring quoting → RFC 4180 double-quote escaping ✓
• parse backslash-escaped quotes → backslash stripped (lossy) ✓
• parse → stringify → parse round-trip → identical output ✓
• parse empty string → {rows: []} ✓
• stringify empty rows → "" ✓
• parse CRLF line endings → handled same as LF ✓