Cortex AI — OpenClaw Plugin

State-of-the-art agentic memory for OpenClaw powered by Cortex AI. Automatically captures conversations, recalls relevant context with knowledge-graph connections, and injects them before every AI turn.

Install

openclaw plugins install @usecortex_ai/openclaw-cortex-ai

Restart OpenClaw after installing.

If you run OpenClaw via the local gateway, restart it too:

openclaw gateway restart

Interactive Onboarding

Run the interactive CLI wizard (recommended):

# Basic onboarding (API key, tenant ID, sub-tenant, ignore term)
openclaw cortex onboard

# Advanced onboarding (all options including recall mode, graph context, etc.)
openclaw cortex onboard --advanced

The wizard guides you through configuration with colored prompts and writes your config to plugins.entries.openclaw-cortex-ai.config inside OpenClaw's settings file.

The path is resolved in the same order OpenClaw itself uses:

1. $OPENCLAW_CONFIG_PATH — if set, used directly
2. $OPENCLAW_STATE_DIR/openclaw.json — if OPENCLAW_STATE_DIR is set
3. $OPENCLAW_HOME/.openclaw/openclaw.json — if OPENCLAW_HOME is set
4. Default: ~/.openclaw/openclaw.json (macOS/Linux) or %USERPROFILE%\.openclaw\openclaw.json (Windows)

No manual adjustment needed — the wizard auto-detects the correct path.

After onboarding, restart the gateway:

openclaw gateway restart

Manual Configuration

If you prefer, you can configure credentials manually.

Two required values:

• API key
• Tenant ID

Environment variables (recommended for secrets):

export CORTEX_OPENCLAW_API_KEY="your-api-key"
export CORTEX_OPENCLAW_TENANT_ID="your-tenant-id"

Or configure directly in OpenClaw's settings file:

• macOS / Linux: ~/.openclaw/openclaw.json
• Windows: %USERPROFILE%\.openclaw\openclaw.json

{
  "plugins": {
    "entries": {
      "openclaw-cortex-ai": {
        "enabled": true,
        "config": {
          "apiKey": "${CORTEX_OPENCLAW_API_KEY}",
          "tenantId": "${CORTEX_OPENCLAW_TENANT_ID}"
        }
      }
    }
  }
}

After changing config, restart the gateway so the plugin reloads:

openclaw gateway restart

Options

| Key | Type | Default | Description | | -------------------- | ----------- | --------------------- | ------------------------------------------------------------------------------ | | subTenantId | string | "cortex-openclaw-plugin" | Sub-tenant for data partitioning within your tenant | | autoRecall | boolean | true | Inject relevant memories before every AI turn | | autoCapture | boolean | true | Store conversation exchanges after every AI turn | | maxRecallResults | number | 10 | Max memory chunks injected into context per turn | | recallMode | string | "fast" | "fast" or "thinking" (deeper personalised recall with graph traversal) | | graphContext | boolean | true | Include knowledge graph relations in recalled context | | ignoreTerm | string | "cortex-ignore" | Messages containing this term are excluded from recall & capture | | debug | boolean | false | Verbose debug logs |

How It Works

• Auto-Recall — Before every AI turn, queries Cortex (/recall/recall_preferences) for relevant memories and injects graph-enriched context (entity paths, chunk relations, extra context).
• Auto-Capture — After every AI turn, the last user/assistant exchange is sent to Cortex (/memories/add_memory) as conversation pairs with infer: true and upsert: true. The session ID is used as source_id so Cortex groups exchanges per session and builds a knowledge graph automatically.

Slash Commands

| Command | Description | | --------------------------- | ------------------------------------- | | /cortex-onboard | Show current configuration status | | /cortex-remember <text> | Save something to Cortex memory | | /cortex-recall <query> | Search memories with relevance scores | | /cortex-list | List all stored user memories | | /cortex-delete <id> | Delete a specific memory by its ID | | /cortex-get <source_id> | Fetch the full content of a source |

AI Tools

| Tool | Description | | ---------------------- | ----------- | | cortex_store | Save the recent conversation history to Cortex as memory | | cortex_search | Search Cortex memories (returns graph-enriched context) | | cortex_list_memories | List all stored user memories (IDs + summaries) | | cortex_get_content | Fetch full content for a specific source_id | | cortex_delete_memory | Delete a memory by memory_id (use only when user explicitly asks) |

CLI

openclaw cortex onboard             # Interactive onboarding wizard
openclaw cortex onboard --advanced  # Advanced onboarding wizard
openclaw cortex search <query>      # Search memories
openclaw cortex list                # List all user memories
openclaw cortex delete <id>         # Delete a memory
openclaw cortex get <source_id>     # Fetch source content
openclaw cortex status              # Show plugin configuration

Troubleshooting

Not configured. Run openclaw cortex onboard

This means the plugin is enabled, but credentials are missing.

Run:

openclaw cortex onboard
openclaw gateway restart

CLI says a command is unknown

Update/restart the gateway so it reloads the plugin:

openclaw gateway restart

Context Injection

Recalled context is injected inside <cortex-context> tags containing:

• Entity Paths — Knowledge graph paths connecting entities relevant to the query
• Context Chunks — Retrieved memory chunks with source titles, graph relations, and linked extra context