Encode and decode base32 (RFC 4648 and Crockford) via @mukundakatta/base32-mcp (npx)

@mukundakatta/base32-mcp v0.1.0 — RFC 4648 & Crockford base32 encoding

Install & run: npm install @mukundakatta/base32-mcp → entry point dist/server.js (ESM, stdio).

1 tool

Key behaviors

• RFC 4648: standard A-Z2-7 alphabet with = padding (e.g. test → ORSXG5A=)
• Crockford: 0-9A-Z (excluding I, L, O, U) with NO padding (e.g. test → EHJQ6X0)
• Perfect round-trips: encode then decode returns exact original
• Case-insensitive decode: lowercase orsxg5a= → test (same as uppercase)
• Missing padding tolerated: ORSXG5A (no =) decodes correctly
• Invalid characters rejected: !!!invalid!!! returns clear error base32 failed: invalid character: !
• Empty string: encode → empty, no error
• Unicode-safe: non-ASCII text (Turkish, accented chars) encodes/decodes correctly

Gotchas

• Single tool `transform` — NOT separate encode/decode tools. The op param controls the operation.
• Same pattern as `encoding-mcp` from the same author (uses op/variant naming)
• Crockford has NO padding — output is shorter than RFC 4648 for same input
• First call ~90ms JIT, subsequent 2-15ms
• Returns JSON {result: "..."} in text content