nelly-elephant-mcp
Search and resume past Claude Code conversations via MCP. Reads local JSONL session files, runs locally, MIT licensed. An elephant never forgets.
README
nelly-elephant-mcp
An elephant never forgets — and neither should your AI.
Search and resume past Claude Code conversations via the Model Context Protocol (MCP).
Install
npm install -g nelly-elephant-mcp
Or install from source:
git clone https://github.com/SpicyRiceCakes/nelly-elephant-mcp.git
cd nelly-elephant-mcp
npm install && npm run build
npm install -g .
Configure
Where Nelly looks
Claude Code stores all session transcripts — from every project — in one place:
~/.claude/projects/
├── -Users-you-project-alpha/ ← sessions from project alpha
│ ├── abc123.jsonl
│ └── def456.jsonl
├── -Users-you-project-beta/ ← sessions from project beta
│ └── ghi789.jsonl
└── ... ← every project you've ever opened
Nelly searches all of them at once. It doesn't matter which project you're in when you ask — she searches everything.
Recommended: Global config
Because Nelly searches across all projects, install her globally so she's available in every session. Add to ~/.claude/mcp_settings.json:
{
"mcpServers": {
"nelly": {
"command": "nelly-elephant-mcp",
"args": []
}
}
}
Alternative: Per-project config
If you only want Nelly in specific projects, add to your project's .mcp.json instead:
{
"mcpServers": {
"nelly": {
"command": "nelly-elephant-mcp",
"args": []
}
}
}
Even in per-project mode, Nelly still searches all sessions. The only difference is whether she's available to call.
Restart Claude Code after configuring. Nelly is ready.
Trunk Modes
Nelly has two modes — just say which one you want:
- 🐘 Long Trunk — the full Nelly experience. She'll talk you through the search, ask follow-up questions, and celebrate when she finds it. An elephant never forgets.
- 🐘 Short Trunk — just the results. No personality, no narration. Date, project, snippet, resume command. Done.
Say "Nelly long trunk" or "Nelly short trunk" to switch anytime.
Tools
| Tool | What it does |
|---|---|
nelly_instructions |
Returns usage guide and system status. |
nelly_search |
Search past sessions by keyword. Supports date range and project filters. |
nelly_context |
Get expanded context from a specific session. |
nelly_recent |
List most recent sessions. |
How it works
Claude Code stores session transcripts as JSONL files in ~/.claude/projects/. Nelly searches these files, extracts human-readable text, and returns matching sessions with claude --resume commands.
You can filter by project name or date range if you want to narrow the search. But by default, Nelly searches everything — that's the point.
The AI uses a "spiral search" pattern — starting broad, trying variations, and asking for hints rather than giving up after one attempt.
Privacy
Elephants have trunks, not phones.
What Nelly does: Reads local files on your machine. Zero network calls. Zero telemetry. Zero data collection. Nelly doesn't send data anywhere. Read the source — it's four files.
What you should know: Nelly runs inside Claude Code. When she finds a matching session, the snippets and project names she returns become part of your current conversation for Claude to process — the same way anything you discuss in a Claude session does. Nelly isn't sending data anywhere extra — it's just that whatever she finds becomes part of the chat, like reading a file or pasting text.
In plain English: Nelly searches locally, but the results become part of your conversation. For sensitive searches, use grep directly in your terminal instead — that stays fully on your machine.
Use Nelly for moments of need, not as a browsing tool. Every search puts past session content into your current conversation. Search when you need to find something specific, not to casually explore your history.
Never use Nelly to search for passwords, API keys, or secrets. If a past session contains a leaked credential, searching for it with Nelly copies it into a second session — doubling the exposure. Use grep directly in your terminal instead:
grep -r "your-search-term" ~/.claude/projects/
That stays fully local. No API, no cloud, no second copy.
Compatibility
| Version | Platform | Status |
|---|---|---|
| v1.0 | Claude Code (macOS, Windows, Linux) | Live |
| v1.a | Codex CLI | PRs welcome |
| v1.b | Gemini CLI | PRs welcome |
| v1.c | Cursor | Research needed |
| v1.d | Windsurf | Research needed |
| v1.e | Cline | Research needed |
Contributing
We built the Claude Code adapter. Want to make one for your tool?
- Fork the repo and look at
src/search.ts— that's the pattern - Adapters need to find session files and parse them for your tool's format
- PRs welcome
Origin
Born from a real need — couldn't find a fix from a previous session. Knew it was somewhere in the conversation history but couldn't remember which one. Searched with grep, found it, realized this should be a tool. Built it that night.
License
MIT
Built by SpicyRiceCakes x Claude
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.