# CogniCore

CogniCore gives agents persistent memory across sessions, reflects on failed attempts, and screens unsafe prompts. It works through an MCP connection.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 81
- Author: cognicore-dev
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/cognicore-dev/cognicore-env
- Updated: 2026-10-21
- HTML: https://affpapa.org/en/nevibecode/mcp/cognicore

## What it does

CogniCore lets an agent retain experience instead of starting each run from scratch. It stores past attempts, recalls related context, and surfaces lessons from repeated failures. Its safety layer can stop prompt injection and jailbreak attempts before they reach the agent.

The hosted MCP server connects to Claude Code, Codex, Cursor, and other compatible clients at `https://mcp.cognicoree.tech/mcp/stream`. A local Python setup needs Python 3.10+ and runs without a vector database or mandatory dependencies. Memory, reflection, and safety do not require an API key.

CogniCore also records decisions for replay and branching, and can export verified memory as a signed bundle. The hosted service uses GitHub sign-in; the local package runs in the Python process.

## Requirements
- Remote server: `https://mcp.cognicoree.tech/mcp/stream`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http cognicore-env https://mcp.cognicoree.tech/mcp/stream
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add cognicore --url https://mcp.cognicoree.tech/mcp/stream
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "cognicore": {
            "url": "https://mcp.cognicoree.tech/mcp/stream"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "cognicore": {
            "type": "http",
            "url": "https://mcp.cognicoree.tech/mcp/stream"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http cognicore https://mcp.cognicoree.tech/mcp/stream
```

## When to use it
- retain agent experience across sessions
- bring earlier attempts into a new run
- turn repeated failures into lessons
- stop prompt injection and jailbreak attempts
- replay a decision and branch from it

## Example prompts
- Save what we learned from this debugging session for the next run.
- Review earlier failed approaches to this bug before suggesting a fix.
- Record this decision so I can replay it and branch from a later step.

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md