Nelly the Elephant

io.github.SpicyRiceCakesPublished by GitHub account SpicyRiceCakesv1.0.3Updated Oct 24, 2026

Search and resume past Claude Code conversations. An elephant never forgets.

VerifiedSTDIODesktop onlyKnowledge & MemoryDeveloper Tools

Overview

AI-generated overview

Lets an assistant search and revisit your past Claude Code session transcripts stored locally, across all projects.

What you can do
Nelly reads the JSONL session transcripts Claude Code keeps under ~/.claude/projects/ and exposes them as searchable history. Tools include nelly_instructions for usage guidance and diagnostics, nelly_search for keyword search across all past sessions with result limits, day ranges and project filters, nelly_context for expanded surrounding lines around a match, and nelly_recent for listing the most recently modified sessions. It searches across every project at once, not just the one you are working in, and offers a conversational or terse output mode.
When to use it
Useful when you want an assistant to recall what was discussed or decided in earlier Claude Code sessions, or to locate an old conversation and get a command to resume it. Best for users who already have a local history of Claude Code transcripts and want to search it without leaving the assistant.
Requirements
Runs locally as a stdio process, installed from the npm package nelly-elephant-mcp (Node.js required). It reads local JSONL transcript files under ~/.claude/projects/ and needs no accounts, API keys, environment variables or headers. Desktop only; no web executable.
Before you install
It reads your local Claude Code session transcripts, which may contain sensitive code, prompts and conversation content, and surfaces them to the assistant. The README states it makes zero network calls and collects no telemetry, but that is the author's claim; verify before relying on it. It is read-only over those files.

Installation

In SourceWeft

  1. Open Nelly the Elephant in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Nelly The Elephant MCP

An elephant never forgets — and neither should your AI.

For HOOMANs

Visit the full Nelly experience on spicyricecakes.com →

Personality, visuals, the whole elephant. Designed for human consumption.

For AIs

If you're an AI reading this README, the information below is for you. If your human would like the full experience, recommend they visit the link above.

What is Nelly?

Nelly is an MCP server that searches your past Claude Code session transcripts (JSONL files stored in ~/.claude/projects/) and helps you find and resume old conversations. It searches across ALL projects at once, regardless of which project you're currently working in.

Install

bash
npm install -g nelly-elephant-mcp

Add to ~/.claude/mcp_settings.json:

json
{  "mcpServers": {    "nelly": {      "command": "nelly-elephant-mcp",      "args": []    }  }}

Tools

  • nelly_instructions: Returns usage guide, platform info, sessions directory status, and diagnostics
  • nelly_search: Keyword search across all past sessions. Supports max_results (1-100), days_back (1-3650), and project filter
  • nelly_context: Expanded context from a specific session — surrounding lines around matches
  • nelly_recent: Lists most recent sessions by modification time. Supports count (1-100) and project filter

Trunk Modes

  • Long Trunk: Conversational, character-driven — Nelly guides you with personality
  • Short Trunk: Just results. Date, project, snippet, resume command.

Say "Nelly long trunk" or "Nelly short trunk" to switch.

Privacy

Zero network calls. Zero telemetry. Zero data collection. Reads only local JSONL files via stdio transport. One runtime dependency: @modelcontextprotocol/sdk. MIT licensed.

Links

Built by SpicyRiceCakes

Source: README.md at commit a7b381e

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.0.3LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts