Metadata-Version: 2.4
Name: novyx-mcp
Version: 2.2.0
Summary: Persistent memory for AI agents. 64 MCP tools for remember, recall, rollback, audit, traces, eval, knowledge graph, and context spaces. Pairs with CLAUDE.md to give Claude Code shared memory across sessions and agents. Works locally (zero config) or with Novyx Cloud.
Author-email: Novyx Labs <blake@novyxlabs.com>
Maintainer-email: Novyx Labs <blake@novyxlabs.com>
License: MIT
Project-URL: Homepage, https://novyxlabs.com
Project-URL: Documentation, https://docs.novyxlabs.com
Keywords: mcp,novyx,memory,ai,agents,claude,claude-code,claude-md,model-context-protocol,multi-agent
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: novyx>=2.0.0
Requires-Dist: mcp>=1.2.0
Provides-Extra: local
Requires-Dist: sentence-transformers>=2.0.0; extra == "local"
Provides-Extra: dev
Requires-Dist: pytest>=7.0.0; extra == "dev"
Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
Dynamic: license-file

<!-- mcp-name: io.github.novyxlabs/novyx-mcp -->
# novyx-mcp

Persistent memory for AI agents. 31 MCP tools for **Claude Desktop**, **Cursor**, and **Claude Code**. Install the server, add a section to your `CLAUDE.md`, and every Claude Code session shares context — turning isolated agents into a coordinated team. Includes draft review before memory is merged, field-level draft diffs, branch/session review, context spaces for multi-agent collaboration, replay for time-travel debugging, and cortex for autonomous memory intelligence.

## Install

```bash
pip install novyx-mcp
```

## Configuration

### Claude Desktop

Add to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "novyx-memory": {
      "command": "python",
      "args": ["-m", "novyx_mcp"],
      "env": {
        "NOVYX_API_KEY": "nram_your_key_here"
      }
    }
  }
}
```

### Cursor

Add to your Cursor MCP settings:

```json
{
  "mcpServers": {
    "novyx-memory": {
      "command": "python",
      "args": ["-m", "novyx_mcp"],
      "env": {
        "NOVYX_API_KEY": "nram_your_key_here"
      }
    }
  }
}
```

### Claude Code

```bash
claude mcp add novyx-memory -- python -m novyx_mcp
```

Set the `NOVYX_API_KEY` environment variable before starting Claude Code. Omit it to use local mode (SQLite, zero config).

### CLAUDE.md Integration

After installing and configuring the MCP server above, add this to your project's `CLAUDE.md` so Claude Code uses Novyx automatically:

```markdown
## Shared Memory (Novyx MCP)
You have access to novyx-mcp tools for shared memory, knowledge graph, audit,
rollback, replay, and context spaces. Use them when relevant. Store decisions
and status at the end of tasks. Check for context from other agents before
starting new work.
```

This turns isolated Claude Code sessions into a coordinated team — each session stores what it learned and checks what other sessions have done before starting work.

## Canonical Workflow: Draft, Review, Merge

This is the highest-signal Novyx workflow for coding agents:

1. The agent learns something important, but uses `draft_memory(..., branch_id="feature-x")` instead of writing directly.
2. Review the whole branch with `memory_branch("feature-x")`.
3. Use `draft_diff` when one draft needs a closer look.
4. Merge the whole branch with `merge_branch("feature-x")`, or reject it with `reject_branch("feature-x")`.

Example:

```text
draft_memory(
  observation="Deploys fail if REDIS_URL is unset in staging",
  tags=["ops", "staging"],
  importance=8,
  branch_id="staging-fixes"
)

memory_branch("staging-fixes")

draft_diff("drf_abc123")

merge_branch("staging-fixes")
```

This keeps agent memory reviewable instead of letting every session write directly into permanent state.

## Available Tools

### Core Memory (18 tools)

| Tool | Description |
|------|-------------|
| `remember` | Store a memory observation with optional tags, importance, context, and TTL |
| `draft_memory` | Create a reviewable draft before writing to canonical memory |
| `memory_drafts` | List open, merged, or rejected drafts |
| `draft_diff` | Show field-level changes before merging a draft |
| `merge_draft` | Merge a reviewed draft into canonical memory |
| `reject_draft` | Reject a draft without changing canonical memory |
| `memory_branch` | Review a whole branch/session of drafts at once |
| `merge_branch` | Merge every open draft in a branch/session |
| `reject_branch` | Reject every open draft in a branch/session |
| `recall` | Search memories semantically using natural language |
| `forget` | Delete a memory by UUID |
| `list_memories` | List stored memories with optional tag filtering |
| `memory_stats` | Get memory statistics (total count, average importance, etc.) |
| `rollback` | Rollback memory to a point in time (supports dry run) |
| `audit` | Get the audit trail of memory operations |
| `link_memories` | Create a directed link between two memories |
| `add_triple` | Add a knowledge graph triple (subject -> predicate -> object) |
| `query_triples` | Query knowledge graph triples with filters |

### Context Spaces (6 tools)

Multi-agent collaboration — shared memory with fine-grained permissions.

| Tool | Description |
|------|-------------|
| `create_space` | Create a shared context space for multi-agent collaboration |
| `list_spaces` | List all context spaces you own or have access to |
| `space_memories` | Search or list memories within a specific context space |
| `update_space` | Update space settings (description, allowed agents, tags) |
| `delete_space` | Delete a context space and disassociate its memories |
| `share_space` | Share a context space with another user by email |

### Replay (4 tools — Pro+)

Time-travel debugging — inspect how memory changed over time.

| Tool | Description |
|------|-------------|
| `replay_timeline` | Get a chronological timeline of memory operations |
| `replay_snapshot` | Reconstruct memory state at a specific point in time |
| `replay_lifecycle` | Trace the full lifecycle of a single memory |
| `replay_diff` | Compare memory state between two points in time |

### Cortex (3 tools — Pro+)

Autonomous memory intelligence — consolidation, reinforcement, and insights.

| Tool | Description |
|------|-------------|
| `cortex_status` | Check cortex configuration and last run stats |
| `cortex_run` | Trigger a cortex cycle (consolidation + reinforcement) |
| `cortex_insights` | Get AI-generated insights from memory patterns (Enterprise) |

## Available Resources

| URI | Description |
|-----|-------------|
| `novyx://memories` | List all stored memories |
| `novyx://memories/{memory_id}` | Get a specific memory by UUID |
| `novyx://stats` | Memory statistics |
| `novyx://usage` | Usage and plan information |
| `novyx://spaces` | List all context spaces |
| `novyx://spaces/{space_id}` | Get a specific context space |

## Available Prompts

| Prompt | Description |
|--------|-------------|
| `memory-context` | Recall relevant memories and format them as context (takes a `query` argument) |
| `session-summary` | List all memories for a session (takes a `session_id` argument) |
| `space-context` | Recall memories from a specific context space (takes `space_id` and `query` arguments) |

## Get an API Key

Sign up at [novyxlabs.com](https://novyxlabs.com) to get your API key. The free tier includes 5,000 memories and 5,000 API calls per month, including the draft-review-merge workflow.

## License

MIT - Copyright 2026 Novyx Labs
