nex（下一步）

Nex - Real-time Organizational Context for AI Agents

Nex shares real-time organizational context with your AI agent: query your context graph, process conversations, receive live insights, and manage the underlying records, schemas, relationships, tasks, and notes.

Setup

1. Get your API key from https://app.nex.ai/settings/developer
2. Add to ~/.openclaw/openclaw.json:

   {
     "skills": {
       "entries": {
         "nex": {
           "enabled": true,
           "env": {
             "NEX_API_KEY": "sk-your_key_here"
           }
         }
       }
     }
   }

Security & Privacy

• All API calls are routed through a validated wrapper script (scripts/nex-api.sh)
• The wrapper validates that all requests go to https://app.nex.ai/api/developers only
• API key is read from $NEX_API_KEY environment variable (never from prompts)
• JSON request bodies are passed via stdin (printf '%s' pipe) to avoid shell injection
• The wrapper uses set -euo pipefail for safe shell execution

IMPORTANT — Safe command construction:

• NEVER interpolate user-supplied text directly into the shell command string
• ALWAYS use printf '%s' '{...}' to pipe JSON via stdin — printf '%s' does not interpret escape sequences or variables in the format argument
• If user input must appear in a JSON body, construct the JSON object using jq: jq -n --arg q "user text" '{query: $q}'
• The examples below use hardcoded JSON for clarity — when building commands with dynamic values, always use the jq construction pattern above

External Endpoints

How to Make API Calls

CRITICAL: The Nex API can take 10-60 seconds to respond. You MUST set timeout: 120 on every exec tool call.

All API calls go through the wrapper script at {baseDir}/scripts/nex-api.sh:

GET request:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/objects",
  "timeout": 120
}

POST with JSON body (pipe body via stdin):

{
  "tool": "exec",
  "command": "printf '%s' '{\"query\":\"What do I know about John?\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/context/ask",
  "timeout": 120
}

GET with jq post-processing (pipe output through jq):

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/insights?last=1h' | jq '[.insights[] | {type, content}]'",
  "timeout": 120
}

SSE stream:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh sse /v1/insights/stream",
  "timeout": 120
}

Handling Large Responses

Nex API responses (especially Insights and List Records) can be 10KB-100KB+. The exec tool may truncate output. You MUST handle this properly.

Pipe output through jq to extract only what you need:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/insights?last=1h' | jq '[.insights[] | {type, content, confidence_level, who: .target.hint}]'",
  "timeout": 120
}

Rules for processing API output:

1. Validate JSON before parsing. If the response doesn't start with { or [, the output may be truncated — retry with a smaller page size or time window.
2. Use jq to keep responses small. Pipe output through jq to extract only the fields you need.
3. Present insights to the user for review. Summarize what was returned and let the user decide which insights to act on.

API Scopes

Each API key has scopes that control access. Request the scopes you need when creating your key at https://app.nex.ai/settings/developer

Choosing the Right API

Before calling an endpoint, decide which approach fits:

Key insight: ProcessText does everything Create/Update Record does, *plus* it extracts relationships, generates insights, and handles ambiguity. Prefer ProcessText when working with conversational or unstructured data. Only use the deterministic Record APIs when you have clean, structured data with known attribute slugs.

Capabilities

Schema Management

Create Object Definition

Create a new custom object type.

Endpoint: POST /v1/objects Scope: object.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"name\":\"Project\",\"name_plural\":\"Projects\",\"slug\":\"project\",\"description\":\"Project tracker\",\"type\":\"custom\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/objects",
  "timeout": 120
}

Get Object Definition

Get a single object definition with its attributes.

Endpoint: GET /v1/objects/{slug} Scope: object.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/objects/project",
  "timeout": 120
}

List Objects

Discover available object types (person, company, etc.) and their attribute schemas. Call this first to learn what fields are available before creating or querying records.

Endpoint: GET /v1/objects Scope: object.read

Query Parameters:

• include_attributes (boolean, optional) -- Set true to include attribute definitions

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/objects?include_attributes=true'",
  "timeout": 120
}

Response:

{
  "data": [
    {
      "id": "123",
      "slug": "person",
      "name": "Person",
      "name_plural": "People",
      "type": "object",
      "description": "A contact or person",
      "attributes": [
        {
          "id": "1",
          "slug": "name",
          "name": "Name",
          "type": "name",
          "options": {
            "is_required": true,
            "is_unique": false,
            "is_multi_value": false
          }
        }
      ]
    }
  ]
}

Update Object Definition

Update an existing object definition.

Endpoint: PATCH /v1/objects/{slug} Scope: object.write

Request body (all fields optional):

{
  "tool": "exec",
  "command": "printf '%s' '{\"name\":\"Updated Project\",\"description\":\"Updated description\"}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/objects/project",
  "timeout": 120
}

Delete Object Definition

Delete an object definition and all its records.

Endpoint: DELETE /v1/objects/{slug} Scope: object.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/objects/project",
  "timeout": 120
}

Create Attribute Definition

Add a new attribute to an object type.

Endpoint: POST /v1/objects/{slug}/attributes Scope: object.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"name\":\"Status\",\"slug\":\"status\",\"type\":\"select\",\"description\":\"Current status\",\"options\":{\"is_required\":true,\"select_options\":[{\"name\":\"Open\"},{\"name\":\"In Progress\"},{\"name\":\"Done\"}]}}' | bash {baseDir}/scripts/nex-api.sh POST /v1/objects/project/attributes",
  "timeout": 120
}

Update Attribute Definition

Update an existing attribute definition.

Endpoint: PATCH /v1/objects/{slug}/attributes/{attr_id} Scope: object.write

Request body (all fields optional):

{
  "tool": "exec",
  "command": "printf '%s' '{\"name\":\"Updated Status\",\"options\":{\"is_required\":false}}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/objects/project/attributes/456",
  "timeout": 120
}

Delete Attribute Definition

Remove an attribute from an object type.

Endpoint: DELETE /v1/objects/{slug}/attributes/{attr_id} Scope: object.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/objects/project/attributes/456",
  "timeout": 120
}

Records

Prefer ProcessText over these endpoints when your input is unstructured text (conversation transcripts, meeting notes, emails). ProcessText automatically creates and updates records, extracts relationships, and generates insights -- all from raw text. Use the endpoints below only when you have clean, structured data with known attribute slugs and values.

Create Record

Create a new record for an object type.

Endpoint: POST /v1/objects/{slug} Scope: record.write

Request body:

• attributes (required) -- Must include name (string or object). Additional fields depend on the object schema.

{
  "tool": "exec",
  "command": "printf '%s' '{\"attributes\":{\"name\":{\"first_name\":\"Jane\",\"last_name\":\"Doe\"},\"email\":\"jane@example.com\",\"company\":\"Acme Corp\"}}' | bash {baseDir}/scripts/nex-api.sh POST /v1/objects/person",
  "timeout": 120
}

Response:

{
  "id": "789",
  "object_id": "123",
  "type": "person",
  "workspace_id": "111",
  "attributes": {
    "name": {"first_name": "Jane", "last_name": "Doe"},
    "email": "jane@example.com",
    "company": "Acme Corp"
  },
  "created_at": "2026-02-11T10:00:00Z",
  "updated_at": "2026-02-11T10:00:00Z"
}

Upsert Record

Create a record if it doesn't exist, or update it if a match is found on the specified attribute.

Endpoint: PUT /v1/objects/{slug} Scope: record.write

Request body:

• attributes (required) -- Must include name when creating
• matching_attribute (required) -- Slug or ID of the attribute to match on (e.g., email)

{
  "tool": "exec",
  "command": "printf '%s' '{\"matching_attribute\":\"email\",\"attributes\":{\"name\":\"Jane Doe\",\"email\":\"jane@example.com\",\"job_title\":\"VP of Sales\"}}' | bash {baseDir}/scripts/nex-api.sh PUT /v1/objects/person",
  "timeout": 120
}

Get Record

Retrieve a specific record by its ID.

Endpoint: GET /v1/records/{record_id} Scope: record.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/records/789",
  "timeout": 120
}

Update Record

Update specific attributes on an existing record. Only the provided attributes are changed.

Endpoint: PATCH /v1/records/{record_id} Scope: record.write

Request body:

• attributes -- Object with the fields to update

{
  "tool": "exec",
  "command": "printf '%s' '{\"attributes\":{\"job_title\":\"CTO\",\"phone\":\"+1-555-0123\"}}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/records/789",
  "timeout": 120
}

Delete Record

Permanently delete a record.

Endpoint: DELETE /v1/records/{record_id} Scope: record.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/records/789",
  "timeout": 120
}

List Records

List records for an object type with optional filtering, sorting, and pagination.

Endpoint: POST /v1/objects/{slug}/records Scope: record.read

Request body:

• attributes -- Which attributes to return: "all", "primary", "none", or a custom object
• limit (integer) -- Number of records to return
• offset (integer) -- Pagination offset
• sort -- Object with attribute (slug) and direction ("asc" or "desc")

{
  "tool": "exec",
  "command": "printf '%s' '{\"attributes\":\"all\",\"limit\":10,\"offset\":0,\"sort\":{\"attribute\":\"updated_at\",\"direction\":\"desc\"}}' | bash {baseDir}/scripts/nex-api.sh POST /v1/objects/person/records",
  "timeout": 120
}

Response:

{
  "data": [
    {
      "id": "789",
      "type": "person",
      "attributes": {"name": "Jane Doe", "email": "jane@example.com"},
      "created_at": "2026-02-11T10:00:00Z",
      "updated_at": "2026-02-11T10:00:00Z"
    }
  ],
  "total": 42,
  "limit": 10,
  "offset": 0
}

Get Record Timeline

Get paginated timeline events for a record (tasks, notes, attribute changes, etc.).

Endpoint: GET /v1/records/{record_id}/timeline Scope: record.read

Query Parameters:

• limit (int) -- Max events (1-100, default: 50)
• cursor (string) -- Pagination cursor from previous response

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/records/1001/timeline?limit=20'",
  "timeout": 120
}

Response:

{
  "data": [
    {
      "id": "5000",
      "resource_type": "task",
      "resource_id": "800",
      "event_type": "created",
      "event_payload": {
        "task": {"id": "800", "title": "Follow up", "priority": "high"}
      },
      "event_timestamp": "2026-02-13T10:00:00Z",
      "created_by": "developer_api"
    }
  ],
  "has_next_page": true,
  "next_cursor": "4999"
}

Resource types: entity, task, note, list_item, attribute Event types: created, updated, deleted, archived

Relationships

Create Relationship Definition

Define a relationship type between two object types.

Endpoint: POST /v1/relationships Scope: relationship.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"type\":\"one_to_many\",\"entity_definition_1_id\":\"123\",\"entity_definition_2_id\":\"456\",\"entity_1_to_2_predicate\":\"has\",\"entity_2_to_1_predicate\":\"belongs to\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/relationships",
  "timeout": 120
}

Response:

{
  "id": "789",
  "type": "one_to_many",
  "entity_definition_1_id": "123",
  "entity_definition_2_id": "456",
  "entity_1_to_2_predicate": "has",
  "entity_2_to_1_predicate": "belongs to",
  "created_at": "2026-02-13T10:00:00Z"
}

List Relationship Definitions

Get all relationship definitions in the workspace.

Endpoint: GET /v1/relationships Scope: relationship.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/relationships",
  "timeout": 120
}

Delete Relationship Definition

Remove a relationship definition.

Endpoint: DELETE /v1/relationships/{id} Scope: relationship.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/relationships/789",
  "timeout": 120
}

Create Relationship Instance

Link two records using an existing relationship definition.

Endpoint: POST /v1/records/{record_id}/relationships Scope: relationship.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"definition_id\":\"789\",\"entity_1_id\":\"1001\",\"entity_2_id\":\"2002\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/records/1001/relationships",
  "timeout": 120
}

Delete Relationship Instance

Remove a relationship between two records.

Endpoint: DELETE /v1/records/{record_id}/relationships/{relationship_id} Scope: relationship.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/records/1001/relationships/5001",
  "timeout": 120
}

Lists

List Object Lists

Get all lists associated with an object type.

Endpoint: GET /v1/objects/{slug}/lists Scope: list.read

Parameters:

• slug (path) -- Object type slug (e.g., person, company)
• include_attributes (query, optional) -- Include attribute definitions

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/objects/person/lists?include_attributes=true'",
  "timeout": 120
}

Response:

{
  "data": [
    {
      "id": "456",
      "slug": "vip-contacts",
      "name": "VIP Contacts",
      "type": "list",
      "attributes": []
    }
  ]
}

Create List

Create a new list under an object type.

Endpoint: POST /v1/objects/{slug}/lists Scope: object.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"name\":\"VIP Contacts\",\"slug\":\"vip-contacts\",\"description\":\"High-value contacts\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/objects/contact/lists",
  "timeout": 120
}

Get List

Get a list definition by ID.

Endpoint: GET /v1/lists/{id} Scope: list.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/lists/300",
  "timeout": 120
}

Delete List

Delete a list definition.

Endpoint: DELETE /v1/lists/{id} Scope: object.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/lists/300",
  "timeout": 120
}

Add List Member

Add an existing record to a list.

Endpoint: POST /v1/lists/{id} Scope: list.member.write

Request body:

• parent_id (required) -- ID of the existing record to add
• attributes (optional) -- List-specific attribute values

{
  "tool": "exec",
  "command": "printf '%s' '{\"parent_id\":\"789\",\"attributes\":{\"status\":\"active\"}}' | bash {baseDir}/scripts/nex-api.sh POST /v1/lists/456",
  "timeout": 120
}

Upsert List Member

Add a record to a list, or update its list-specific attributes if already a member.

Endpoint: PUT /v1/lists/{id} Scope: list.member.write

{
  "tool": "exec",
  "command": "printf '%s' '{\"parent_id\":\"789\",\"attributes\":{\"priority\":\"high\"}}' | bash {baseDir}/scripts/nex-api.sh PUT /v1/lists/456",
  "timeout": 120
}

List Records in a List

Get paginated records from a specific list.

Endpoint: POST /v1/lists/{id}/records Scope: list.member.read

Request body: Same as List Records (attributes, limit, offset, sort)

{
  "tool": "exec",
  "command": "printf '%s' '{\"attributes\":\"all\",\"limit\":20}' | bash {baseDir}/scripts/nex-api.sh POST /v1/lists/456/records",
  "timeout": 120
}

Update List Record

Update list-specific attributes for a record within a list.

Endpoint: PATCH /v1/lists/{id}/records/{record_id} Scope: list.member.write

{
  "tool": "exec",
  "command": "printf '%s' '{\"attributes\":{\"status\":\"closed-won\"}}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/lists/456/records/789",
  "timeout": 120
}

Delete List Record

Remove a record from a list.

Endpoint: DELETE /v1/lists/{id}/records/{record_id} Scope: record.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/lists/300/records/4001",
  "timeout": 120
}

Tasks

Create Task

Create a new task, optionally linked to records and assigned to users.

Endpoint: POST /v1/tasks Scope: task.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"title\":\"Follow up with client\",\"description\":\"Discuss contract renewal\",\"priority\":\"high\",\"due_date\":\"2026-03-01T09:00:00Z\",\"entity_ids\":[\"1001\",\"1002\"],\"assignee_ids\":[\"50\"]}' | bash {baseDir}/scripts/nex-api.sh POST /v1/tasks",
  "timeout": 120
}

Response:

{
  "id": "800",
  "title": "Follow up with client",
  "description": "Discuss contract renewal",
  "priority": "high",
  "due_date": "2026-03-01T09:00:00Z",
  "assignee_ids": ["50"],
  "entity_ids": ["1001", "1002"],
  "created_by": "developer_api",
  "created_at": "2026-02-13T10:00:00Z"
}

List Tasks

List tasks with optional filtering.

Endpoint: GET /v1/tasks Scope: task.read

Query Parameters:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/tasks?entity_id=1001&is_completed=false&limit=20'",
  "timeout": 120
}

Response:

{
  "data": [],
  "has_more": true,
  "total": 47,
  "next_offset": 20
}

Get Task

Get a single task by ID.

Endpoint: GET /v1/tasks/{task_id} Scope: task.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/tasks/800",
  "timeout": 120
}

Update Task

Update a task's fields. All fields are optional.

Endpoint: PATCH /v1/tasks/{task_id} Scope: task.write

Request body (all fields optional):

{
  "tool": "exec",
  "command": "printf '%s' '{\"priority\":\"urgent\",\"is_completed\":true}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/tasks/800",
  "timeout": 120
}

Delete Task

Archive a task (soft delete).

Endpoint: DELETE /v1/tasks/{task_id} Scope: task.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/tasks/800",
  "timeout": 120
}

Notes

Create Note

Create a new note, optionally linked to a record.

Endpoint: POST /v1/notes Scope: note.write

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"title\":\"Meeting notes\",\"content\":\"Discussed Q3 roadmap...\",\"entity_id\":\"1001\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/notes",
  "timeout": 120
}

Response:

{
  "id": "900",
  "title": "Meeting notes",
  "content": "Discussed Q3 roadmap...",
  "entity_id": "1001",
  "created_by": "developer_api",
  "created_at": "2026-02-13T10:00:00Z"
}

List Notes

List notes, optionally filtered by associated record.

Endpoint: GET /v1/notes Scope: note.read

Query Parameters:

Response capped at 200 notes max.

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/notes?entity_id=1001'",
  "timeout": 120
}

Get Note

Get a single note by ID.

Endpoint: GET /v1/notes/{note_id} Scope: note.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/notes/900",
  "timeout": 120
}

Update Note

Update a note's fields.

Endpoint: PATCH /v1/notes/{note_id} Scope: note.write

Request body (all fields optional):

{
  "tool": "exec",
  "command": "printf '%s' '{\"title\":\"Updated meeting notes\",\"content\":\"Added action items...\"}' | bash {baseDir}/scripts/nex-api.sh PATCH /v1/notes/900",
  "timeout": 120
}

Delete Note

Archive a note (soft delete).

Endpoint: DELETE /v1/notes/{note_id} Scope: note.write

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh DELETE /v1/notes/900",
  "timeout": 120
}

Search

Search Records

Search records by name across all object types.

Endpoint: POST /v1/search Scope: record.read

Request body:

{
  "tool": "exec",
  "command": "printf '%s' '{\"query\":\"john doe\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/search",
  "timeout": 120
}

Response:

{
  "results": {
    "person": [
      {
        "id": "1001",
        "primary_value": "John Doe",
        "matched_value": "John Doe",
        "score": 0.95,
        "entity_definition_id": "10"
      }
    ],
    "company": [
      {
        "id": "2001",
        "primary_value": "Doe Industries",
        "matched_value": "Doe Industries",
        "score": 0.72,
        "entity_definition_id": "11"
      }
    ]
  },
  "errored_search_types": []
}

Results are grouped by object type (e.g., "person", "company", "deal"). If some search types fail, partial results are returned with errored_search_types listing which types had errors.

Context & AI

Query Context (Ask API)

Use this when you need to recall information about contacts, companies, or relationships.

Endpoint: POST /v1/context/ask Scope: record.read

{
  "tool": "exec",
  "command": "printf '%s' '{\"query\":\"What do I know about John Smith?\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/context/ask",
  "timeout": 120
}

Response:

{
  "answer": "John Smith is a VP of Sales at Acme Corp...",
  "entities_considered": [
    {"id": 123, "name": "John Smith", "type": "contact"}
  ],
  "signals_used": [
    {"id": 456, "content": "Met at conference last month"}
  ],
  "metadata": {
    "query_type": "entity_specific"
  }
}

Example queries:

• "Who are my most engaged contacts this week?"
• "What companies are we working with in the healthcare sector?"
• "What was discussed in my last meeting with Sarah?"

Add Context (ProcessText API)

Use this to ingest new information from conversations, meeting notes, or other text.

Endpoint: POST /v1/context/text Scope: record.write

Request body:

• content (required) -- The text to process
• context (optional) -- Additional context about the text (e.g., "Sales call notes")

{
  "tool": "exec",
  "command": "printf '%s' '{\"content\":\"Had a great call with John Smith from Acme Corp.\",\"context\":\"Sales call notes\"}' | bash {baseDir}/scripts/nex-api.sh POST /v1/context/text",
  "timeout": 120
}

Response:

{
  "artifact_id": "abc123"
}

After calling ProcessText, use Get Artifact Status to check processing results.

Get Artifact Status

Check the processing status and results after calling ProcessText.

Endpoint: GET /v1/context/artifacts/{artifact_id} Scope: record.read

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET /v1/context/artifacts/abc123",
  "timeout": 120
}

Response:

{
  "operation_id": 48066188026052610,
  "status": "completed",
  "result": {
    "entities_extracted": [
      {"name": "John Smith", "type": "PERSON", "action": "created"},
      {"name": "Acme Corp", "type": "COMPANY", "action": "updated"}
    ],
    "entities_created": 1,
    "entities_updated": 1,
    "relationships": 1,
    "insights": [
      {"content": "Acme Corp expanding to APAC", "confidence": 0.85}
    ],
    "tasks": []
  },
  "created_at": "2026-02-05T10:30:00Z",
  "completed_at": "2026-02-05T10:30:15Z"
}

Status values: pending, processing, completed, failed

Typical workflow:

1. Call ProcessText -> get artifact_id
2. Poll Get Artifact Status every 2-5 seconds
3. Stop when status is completed or failed
4. Report the extracted entities and insights to the user

Create AI List Job

Use natural language to search your context graph and generate a curated list of contacts or companies. Use this when you know the object type (contact or company) and want AI to filter and rank matches. If you're unsure which object type applies or the question is open-ended, use the Ask API instead.

Endpoint: POST /v1/context/list/jobs Scope: list.member.write

Request body:

• query (required) -- Natural language search query
• object_type (optional) -- "contact" or "company" (default: "contact")
• limit (optional) -- Number of results (default: 50, max: 100)
• include_attributes (optional) -- Include all entity attribute values (default: false)

{
  "tool": "exec",
  "command": "printf '%s' '{\"query\":\"high priority contacts in enterprise deals\",\"object_type\":\"contact\",\"limit\":20,\"include_attributes\":true}' | bash {baseDir}/scripts/nex-api.sh POST /v1/context/list/jobs",
  "timeout": 120
}

Response:

{
  "message": {
    "job_id": "12345678901234567",
    "status": "pending"
  }
}

Get AI List Job Status

Check status and results of an AI list generation job. Poll until status is completed or failed.

Endpoint: GET /v1/context/list/jobs/{job_id} Scope: list.member.read

Query Parameters:

• include_attributes (boolean, optional) -- Include full attributes for each entity

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/context/list/jobs/12345678901234567?include_attributes=true'",
  "timeout": 120
}

Response (completed):

{
  "message": {
    "job_id": "12345678901234567",
    "status": "completed",
    "created_at": "2026-02-11T10:00:00Z",
    "count": 5,
    "query_interpretation": "Finding contacts involved in enterprise-level deals marked as high priority",
    "entities": [
      {
        "id": "789",
        "name": "Jane Doe",
        "type": "contact",
        "reason": "Lead on $500K enterprise deal with Acme Corp, marked high priority",
        "highlights": [
          "Contract negotiation in progress",
          "Budget approved Q1 2026",
          "Executive sponsor confirmed"
        ],
        "attributes": {}
      }
    ]
  }
}

Status values: pending, processing, completed, failed

Typical workflow:

1. Create job with natural language query -> get job_id
2. Poll Get AI List Job Status every 2-5 seconds
3. Stop when status is completed or failed
4. Present the matched entities with reasons and highlights

Insights

Get Insights (REST)

Query insights by time window. Supports two modes.

Endpoint: GET /v1/insights Scope: insight.stream

Query Parameters:

• last -- Duration window, e.g., 30m, 2h, 1h30m
• from + to -- UTC time range in RFC3339 format
• limit (optional) -- Max results (default: 20, max: 100)

If neither last nor from/to is specified, returns the most recent insights (default 20).

Last 30 minutes:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/insights?last=30m'",
  "timeout": 120
}

Between two dates:

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/insights?from=2026-02-07T00:00:00Z&to=2026-02-08T00:00:00Z'",
  "timeout": 120
}

Recommended: Pipe through jq to extract a summary (responses can be 10-100KB+):

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh GET '/v1/insights?last=1h' | jq '{insight_count: (.insights | length), insights: [.insights[] | {type, content, confidence_level, who: .target.hint, entity_type: .target.entity_type}]}'",
  "timeout": 120
}

When to use:

• When polling periodically instead of maintaining SSE connection
• To get current insight state on startup
• As fallback when SSE connection drops
• To review insights from a specific time period

Notable insight categories (summarize for user review):

• New contacts or companies
• Opportunities, risks, or milestones
• Meetings scheduled or occurred
• Relationship changes
• High-confidence insights (confidence_level of "high" or "very_high")

Real-time Insight Stream (SSE)

Receive insights as they are discovered in real time.

Endpoint: GET /v1/insights/stream Scope: insight.stream

{
  "tool": "exec",
  "command": "bash {baseDir}/scripts/nex-api.sh sse /v1/insights/stream",
  "timeout": 120
}

Connection behavior:

• Server sends : connected workspace_id=... token_id=... on connection
• Recent insights are replayed immediately via insight.replay events (up to 20)
• Keepalive comments (: keepalive) sent every 30 seconds
• Real-time events arrive as: `event: insight.batch.created\

data: {...}\ \ `

Event payload structure:

{
  "workspace": {
    "name": "Acme Corp",
    "slug": "acme",
    "business_info": {"name": "Acme Corp", "domain": "acme.com"}
  },
  "insights": [{
    "type": "opportunity",
    "type_description": "A potential business opportunity",
    "content": "Budget approval expected next quarter",
    "confidence": 0.85,
    "confidence_level": "high",
    "target": {
      "type": "entity",
      "entity_type": "person",
      "hint": "John Smith",
      "signals": [{"type": "email", "value": "john@acme.com"}]
    },
    "evidence": [{
      "excerpt": "We should have budget approval by Q2",
      "artifact": {"type": "email", "subject": "RE: Proposal"}
    }]
  }],
  "insight_count": 1,
  "emitted_at": "2026-02-05T10:30:00Z"
}

Insight types: opportunity, risk, relationship, preference, milestone, activity, characteristic, role_detail

When to use: Keep the SSE connection open in the background during active conversations. For one-off queries, use the Ask API instead.

Error Handling

When to Use Nex

Good use cases:

• Before responding to a message, query for context about the person
• After a conversation, process the transcript to update the context graph
• When asked about relationships or history with contacts/companies
• Creating or updating records from conversation context
• Building targeted lists from your contact database
• Looking up record details before a meeting
• Creating tasks and notes to track follow-ups
• Searching across all record types to find specific people or companies
• Defining custom object schemas and relationships for your workspace

Not for:

• General knowledge questions (use web search)
• Real-time calendar/scheduling (use calendar tools)
• Bulk data entry that requires the full Nex UI