MCP Tools Reference

This page documents every MCP tool exposed by Mimir AIP, organized by category.

All tools return JSON responses. Long-running operations (training, pipeline execution, digital twin sync) return task IDs; use wait_for_task or get_work_task to poll for completion.

Projects (8 tools)

Projects are the top-level organizational unit in Mimir. Each project groups pipelines, ontologies, ML models, digital twins, and storage configurations.

list_projects

List all projects in the platform. Parameters:

status (optional): Filter by status (active, archived, draft)

Example:

{
  "status": "active"
}

Returns: Array of project objects with id, name, description, version, status, tags, created_at, updated_at.

get_project

Get details of a specific project. Parameters:

id (required): Project ID

Example:

{
  "id": "proj_abc123"
}

Returns: Project object with full details including associated component IDs.

create_project

Create a new project. Parameters:

name (required): Project name (3-50 alphanumeric characters, hyphens, underscores)
description (optional): Human-readable description
version (optional): Semantic version string (e.g., 1.0.0)
tags (optional): Comma-separated list of tags (e.g., production,ml,iot)

Example:

{
  "name": "SmartFactory",
  "description": "IoT data aggregation and ML for factory automation",
  "version": "1.0.0",
  "tags": "iot,ml,production"
}

Returns: Created project object.

update_project

Update an existing project’s metadata. Parameters:

id (required): Project ID
description (optional): New description
version (optional): New version (e.g., 2.0.0)
status (optional): New status (active, archived, draft)
tags (optional): Comma-separated replacement tag list

Example:

{
  "id": "proj_abc123",
  "version": "1.1.0",
  "tags": "iot,ml,production,v1.1"
}

Returns: Updated project object.

delete_project

Delete (archive) a project. Parameters:

id (required): Project ID

Example:

{
  "id": "proj_abc123"
}

Returns: {"success": true}

add_project_component

Associate a component with a project. Parameters:

project_id (required): Project ID
component_type (required): One of pipeline, ontology, ml_model, digital_twin, storage
component_id (required): ID of the component to associate

Example:

{
  "project_id": "proj_abc123",
  "component_type": "pipeline",
  "component_id": "pipe_xyz789"
}

Returns: {"success": true}

remove_project_component

Remove a component association from a project. Parameters:

project_id (required): Project ID
component_type (required): One of pipeline, ontology, ml_model, digital_twin, storage
component_id (required): ID of the component to disassociate

Example:

{
  "project_id": "proj_abc123",
  "component_type": "pipeline",
  "component_id": "pipe_xyz789"
}

Returns: {"success": true}

clone_project

Clone an existing project under a new name. Parameters:

id (required): Source project ID
new_name (required): Name for the cloned project

Example:

{
  "id": "proj_abc123",
  "new_name": "SmartFactory-Dev"
}

Returns: Cloned project object with a new ID.

Pipelines (6 tools)

Pipelines define ordered sequences of data processing steps (ingestion → processing → output).

list_pipelines

List pipelines, optionally filtered by project. Parameters:

project_id (optional): Filter by project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of pipeline objects.

get_pipeline

Get details of a specific pipeline. Parameters:

id (required): Pipeline ID

Example:

{
  "id": "pipe_xyz789"
}

Returns: Pipeline object with full step definitions.

create_pipeline

Create a new pipeline. Parameters:

project_id (required): Project ID
name (required): Pipeline name
type (required): One of ingestion, processing, output
steps (required): JSON array of pipeline steps
description (optional): Pipeline description

Step Schema:

{
  "name": "step1",
  "plugin": "default",
  "action": "transform",
  "parameters": {}
}

Example:

{
  "project_id": "proj_abc123",
  "name": "IoT-Ingestion",
  "type": "ingestion",
  "steps": [
    {
      "name": "fetch",
      "plugin": "http",
      "action": "GET",
      "parameters": {"url": "https://api.example.com/data"}
    },
    {
      "name": "transform",
      "plugin": "json",
      "action": "parse",
      "parameters": {}
    }
  ],
  "description": "Fetch and parse IoT data from external API"
}

Returns: Created pipeline object.

update_pipeline

Update an existing pipeline. Parameters:

id (required): Pipeline ID
description (optional): New description
steps (optional): Replacement JSON array of steps
status (optional): New status (active, inactive, archived)

Example:

{
  "id": "pipe_xyz789",
  "status": "active"
}

Returns: Updated pipeline object.

delete_pipeline

Delete a pipeline. Parameters:

id (required): Pipeline ID

Example:

{
  "id": "pipe_xyz789"
}

Returns: {"success": true}

execute_pipeline

Enqueue a pipeline for asynchronous execution. Parameters:

id (required): Pipeline ID

Example:

{
  "id": "pipe_xyz789"
}

Returns:

{
  "task_id": "task_123",
  "pipeline_id": "pipe_xyz789",
  "status": "queued"
}

Use wait_for_task or get_work_task to poll for completion.

Schedules (5 tools)

Schedules are cron-based triggers that enqueue pipelines on a recurring basis.

list_schedules

List schedules, optionally filtered by project. Parameters:

project_id (optional): Filter by project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of schedule objects.

get_schedule

Get details of a specific schedule. Parameters:

id (required): Schedule ID

Example:

{
  "id": "sched_456"
}

Returns: Schedule object with cron expression and associated pipeline IDs.

create_schedule

Create a new cron-based schedule. Parameters:

project_id (required): Project ID
name (required): Schedule name
cron_schedule (required): Cron expression (e.g., "0 * * * *" for hourly)
pipeline_ids (required): Comma-separated list of pipeline IDs to trigger
enabled (optional): Enable immediately (default true)

Example:

{
  "project_id": "proj_abc123",
  "name": "Hourly-Ingestion",
  "cron_schedule": "0 * * * *",
  "pipeline_ids": "pipe_xyz789,pipe_abc456",
  "enabled": "true"
}

Returns: Created schedule object.

update_schedule

Update an existing schedule. Parameters:

id (required): Schedule ID
name (optional): New name
cron_schedule (optional): New cron expression
pipeline_ids (optional): Comma-separated replacement list of pipeline IDs
enabled (optional): Enable/disable (true or false)

Example:

{
  "id": "sched_456",
  "enabled": "false"
}

Returns: Updated schedule object.

delete_schedule

Delete a schedule. Parameters:

id (required): Schedule ID

Example:

{
  "id": "sched_456"
}

Returns: {"success": true}

ML Models (7 tools)

ML models are trained on data from storage backends and used for inference. Supported types: decision tree, random forest, regression, neural network.

list_ml_models

List ML models for a project. Parameters:

project_id (required): Project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of ML model objects.

get_ml_model

Get details of a specific ML model. Parameters:

id (required): ML model ID

Example:

{
  "id": "model_789"
}

Returns: ML model object with training config and status.

create_ml_model

Create a new ML model definition. Parameters:

project_id (required): Project ID
ontology_id (required): Ontology ID that defines the model’s domain
name (required): Model name
type (required): One of decision_tree, random_forest, regression, neural_network
description (optional): Model description
config (optional): JSON training config

Config Schema:

{
  "train_test_split": 0.8,
  "random_seed": 42,
  "max_depth": 10,
  "n_estimators": 100,
  "learning_rate": 0.01
}

Example:

{
  "project_id": "proj_abc123",
  "ontology_id": "onto_xyz",
  "name": "Anomaly-Detector",
  "type": "random_forest",
  "config": "{\"max_depth\":10,\"n_estimators\":100}"
}

Returns: Created ML model object.

update_ml_model

Update an existing ML model’s metadata. Parameters:

id (required): ML model ID
name (optional): New name
description (optional): New description
status (optional): New status (created, training, trained, failed, deprecated)

Example:

{
  "id": "model_789",
  "status": "deprecated"
}

Returns: Updated ML model object.

delete_ml_model

Delete an ML model. Parameters:

id (required): ML model ID

Example:

{
  "id": "model_789"
}

Returns: {"success": true}

train_ml_model

Start asynchronous training for an ML model. Parameters:

model_id (required): ML model ID
storage_ids (optional): Comma-separated list of storage config IDs to use as training data

Example:

{
  "model_id": "model_789",
  "storage_ids": "storage_111,storage_222"
}

Returns: ML model object with status updated to training. Training runs asynchronously as a worker job.

run_inference

Enqueue an ML inference job. Parameters:

model_id (required): Trained ML model ID
storage_id (required): Storage config ID containing data to run inference on

Example:

{
  "model_id": "model_789",
  "storage_id": "storage_111"
}

Returns:

{
  "task_id": "task_456",
  "model_id": "model_789",
  "storage_id": "storage_111",
  "status": "queued"
}

Use wait_for_task or get_work_task to poll for completion.

recommend_model

Recommend the best ML model type for a project based on its ontology and data. Parameters:

project_id (required): Project ID
ontology_id (required): Ontology ID describing the data domain

Example:

{
  "project_id": "proj_abc123",
  "ontology_id": "onto_xyz"
}

Returns:

{
  "recommended_type": "random_forest",
  "confidence": 0.85,
  "reasoning": "Data has categorical and numerical features; random forest performs well on mixed data."
}

Digital Twins (7 tools)

Digital twins are live in-memory entity graphs initialized from an ontology and synchronized from storage. Queryable via SPARQL.

list_digital_twins

List digital twins, optionally filtered by project. Parameters:

project_id (optional): Filter by project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of digital twin objects.

get_digital_twin

Get details of a specific digital twin. Parameters:

id (required): Digital twin ID

Example:

{
  "id": "dt_999"
}

Returns: Digital twin object with ontology ID, status, and entity count.

create_digital_twin

Create a new digital twin. Parameters:

project_id (required): Project ID
ontology_id (required): Ontology ID that defines the twin’s entity model
name (required): Digital twin name
description (optional): Description

Example:

{
  "project_id": "proj_abc123",
  "ontology_id": "onto_xyz",
  "name": "Factory-Twin",
  "description": "Live digital twin of factory floor sensors and machines"
}

Returns: Created digital twin object.

update_digital_twin

Update an existing digital twin’s metadata. Parameters:

id (required): Digital twin ID
name (optional): New name
description (optional): New description
status (optional): New status (active, inactive, archived)

Example:

{
  "id": "dt_999",
  "status": "active"
}

Returns: Updated digital twin object.

delete_digital_twin

Delete a digital twin. Parameters:

id (required): Digital twin ID

Example:

{
  "id": "dt_999"
}

Returns: {"success": true}

sync_digital_twin

Enqueue a digital twin sync job to update entities from storage. Parameters:

id (required): Digital twin ID

Example:

{
  "id": "dt_999"
}

Returns:

{
  "task_id": "task_789",
  "digital_twin_id": "dt_999",
  "status": "queued"
}

Use wait_for_task or get_work_task to poll for completion.

query_digital_twin

Execute a SPARQL query against a digital twin’s entity graph. Parameters:

id (required): Digital twin ID
sparql_query (required): SPARQL query string
limit (optional): Maximum number of results (default 100)

Example:

{
  "id": "dt_999",
  "sparql_query": "SELECT ?sensor ?temp WHERE { ?sensor a :Sensor ; :temperature ?temp . FILTER(?temp > 80) } LIMIT 50",
  "limit": 50
}

Returns:

{
  "results": [
    {"sensor": "sensor_001", "temp": 85.3},
    {"sensor": "sensor_042", "temp": 92.1}
  ],
  "count": 2
}

Ontologies (6 tools)

Ontologies define entity types, properties, and relationships in OWL/Turtle format.

list_ontologies

List ontologies for a project. Parameters:

project_id (required): Project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of ontology objects.

get_ontology

Get details of a specific ontology. Parameters:

id (required): Ontology ID

Example:

{
  "id": "onto_xyz"
}

Returns: Ontology object with full OWL/Turtle content.

create_ontology

Create a new ontology. Parameters:

project_id (required): Project ID
name (required): Ontology name
content (required): OWL/Turtle ontology content as a string
description (optional): Description
version (optional): Version string (default 1.0.0)

Example:

{
  "project_id": "proj_abc123",
  "name": "IoT-Ontology",
  "content": "@prefix : <http://example.com/ont#> . :Sensor a owl:Class . :temperature a owl:DatatypeProperty .",
  "version": "1.0.0"
}

Returns: Created ontology object.

update_ontology

Update an existing ontology. Parameters:

id (required): Ontology ID
name (optional): New name
description (optional): New description
version (optional): New version (e.g., 2.0.0)
content (optional): Replacement OWL/Turtle content
status (optional): New status (active, deprecated)

Example:

{
  "id": "onto_xyz",
  "version": "1.1.0"
}

Returns: Updated ontology object.

delete_ontology

Delete an ontology. Parameters:

id (required): Ontology ID

Example:

{
  "id": "onto_xyz"
}

Returns: {"success": true}

generate_ontology_from_text

Generate an OWL ontology by extracting entity types and relationships from a text description. Parameters:

project_id (required): Project ID
name (required): Name for the generated ontology
text (required): Domain description text

Example:

{
  "project_id": "proj_abc123",
  "name": "Factory-Ontology",
  "text": "A factory has machines. Each machine has sensors that measure temperature and pressure. Sensors send readings to a central database."
}

Returns:

{
  "ontology": { /* ontology object */ },
  "extraction_summary": {
    "entities_count": 4,
    "relationships_count": 2
  }
}

extract_and_generate_ontology

Extract entities from storage backends and generate an OWL ontology. Diffs against any existing active ontology and flags for review if changes are detected. Parameters:

project_id (required): Project ID
storage_ids (required): Comma-separated list of storage config IDs
ontology_name (required): Name for the generated ontology
include_structured (optional): Include structured data extraction (default true)
include_unstructured (optional): Include unstructured/text extraction (default true)

Example:

{
  "project_id": "proj_abc123",
  "storage_ids": "storage_111,storage_222",
  "ontology_name": "Extracted-Ontology"
}

Returns:

{
  "ontology": { /* ontology object */ },
  "extraction_summary": {
    "entities_count": 12,
    "relationships_count": 8
  },
  "ontology_diff": { /* diff object if changes detected */ }
}

extract_from_storage

Extract entities and relationships from one or more storage backends without generating an ontology. Parameters:

project_id (required): Project ID
storage_ids (required): Comma-separated list of storage config IDs
include_structured (optional): Include structured data extraction (default true)
include_unstructured (optional): Include unstructured/text extraction (default true)

Example:

{
  "project_id": "proj_abc123",
  "storage_ids": "storage_111"
}

Returns:

{
  "entities": [ /* array of extracted entities */ ],
  "relationships": [ /* array of extracted relationships */ ],
  "source": "storage"
}

Storage (10 tools)

Storage tools manage backend configurations and CIR (Common Internal Representation) data operations.

list_storage_configs

List storage configurations for a project. Parameters:

project_id (required): Project ID

Example:

{
  "project_id": "proj_abc123"
}

Returns: Array of storage config objects.

get_storage_config

Get a specific storage configuration. Parameters:

id (required): Storage config ID

Example:

{
  "id": "storage_111"
}

Returns: Storage config object with plugin type and configuration.

create_storage_config

Create a new storage configuration. Parameters:

project_id (required): Project ID
type (required): Storage plugin type (filesystem, postgresql, mysql, mongodb, s3, redis, elasticsearch, neo4j)
config (required): JSON object with plugin-specific config

Config Examples: Filesystem:

{"path": "/data/project1"}

PostgreSQL:

{"connection_string": "postgresql://user:pass@localhost:5432/dbname"}

S3:

{"bucket": "my-bucket", "region": "us-east-1", "access_key": "...", "secret_key": "..."}

Example:

{
  "project_id": "proj_abc123",
  "type": "postgresql",
  "config": "{\"connection_string\":\"postgresql://user:pass@localhost:5432/iot\"}"
}

Returns: Created storage config object.

update_storage_config

Update a storage configuration. Parameters:

id (required): Storage config ID
config (optional): JSON object with updated plugin-specific config
active (optional): Set active state (true or false)

Example:

{
  "id": "storage_111",
  "active": "false"
}

Returns: {"success": true}

delete_storage_config

Delete a storage configuration. Parameters:

id (required): Storage config ID

Example:

{
  "id": "storage_111"
}

Returns: {"success": true}

store_data

Store one or more CIR (Common Internal Representation) records. Parameters:

storage_id (required): Storage config ID
data (required): JSON array of CIR objects

CIR Schema:

{
  "version": "1.0",
  "source": {
    "type": "api",
    "uri": "manual",
    "timestamp": "2024-01-01T00:00:00Z",
    "format": "json"
  },
  "data": {
    "entity_type": "Sensor",
    "id": "sensor_001",
    "temperature": 75.2
  },
  "metadata": {
    "project_id": "proj_abc123"
  }
}

Example:

{
  "storage_id": "storage_111",
  "data": "[{\"version\":\"1.0\",\"source\":{\"type\":\"api\",\"uri\":\"manual\",\"timestamp\":\"2024-01-01T00:00:00Z\",\"format\":\"json\"},\"data\":{\"key\":\"value\"},\"metadata\":{}}]"
}

Returns:

{
  "stored": 1,
  "total": 1
}

retrieve_data

Retrieve CIR records from a storage backend. Parameters:

storage_id (required): Storage config ID
entity_type (optional): Filter by entity type
limit (optional): Maximum number of records (default 100)

Example:

{
  "storage_id": "storage_111",
  "entity_type": "Sensor",
  "limit": 50
}

Returns:

{
  "records": [ /* array of CIR objects */ ],
  "count": 50
}

update_data

Update CIR records matching a query. Parameters:

storage_id (required): Storage config ID
query (required): JSON CIRQuery to select records
updates (required): JSON CIRUpdate with fields to set

Example:

{
  "storage_id": "storage_111",
  "query": "{\"entity_type\":\"Sensor\",\"limit\":100}",
  "updates": "{\"set_fields\":{\"status\":\"inactive\"}}"
}

Returns:

{
  "updated": 12
}

delete_data

Delete CIR records matching a query. Parameters:

storage_id (required): Storage config ID
query (required): JSON CIRQuery to select records for deletion

Example:

{
  "storage_id": "storage_111",
  "query": "{\"entity_type\":\"Sensor\"}"
}

Returns:

{
  "deleted": 25
}

storage_health

Check whether a storage backend is reachable and healthy. Parameters:

storage_id (required): Storage config ID

Example:

{
  "storage_id": "storage_111"
}

Returns:

{
  "storage_id": "storage_111",
  "healthy": true
}

Or if unhealthy:

{
  "storage_id": "storage_111",
  "healthy": false,
  "error": "connection refused"
}

Tasks (3 tools)

Work tasks represent asynchronous jobs (pipeline execution, ML training/inference, digital twin sync).

list_work_tasks

List work tasks in the queue. Parameters:

status (optional): Filter by status (queued, scheduled, spawned, executing, completed, failed, timeout, cancelled)
type (optional): Filter by type (pipeline_execution, ml_training, ml_inference, digital_twin_update)

Example:

{
  "status": "executing",
  "type": "ml_training"
}

Returns: Array of work task objects.

get_work_task

Get the current state of a specific work task. Parameters:

id (required): Work task ID

Example:

{
  "id": "task_123"
}

Returns: Work task object with status, progress, and result (if completed).

wait_for_task

Poll a work task until it reaches a terminal state (completed, failed, timeout, cancelled) or the timeout expires. Parameters:

id (required): Work task ID
timeout_seconds (optional): Maximum seconds to wait (default 300, max 600)

Example:

{
  "id": "task_123",
  "timeout_seconds": 600
}

Returns:

If terminal state reached: Work task object with final status
If timeout: {"task": {...}, "timed_out": true, "message": "task did not reach a terminal state within the timeout"}

System (1 tool)

health_check

Check the health status of the Mimir AIP platform. Parameters: None Example:

{}

Returns:

{
  "status": "healthy",
  "platform": "Mimir AIP",
  "version": "1.0.0"
}

Common Patterns

Chaining Operations

Many workflows require chaining multiple tool calls:

create_project → get project_id
create_storage_config → get storage_id
extract_and_generate_ontology → get ontology_id
create_ml_model → get model_id
train_ml_model → get task_id
wait_for_task → wait for training completion

Handling Async Operations

Tools that return task_id are asynchronous:

execute_pipeline
train_ml_model
run_inference
sync_digital_twin

Use wait_for_task or poll get_work_task to track progress:

1. execute_pipeline → {"task_id": "task_123"}
2. wait_for_task({"id": "task_123", "timeout_seconds": 300})
   → {"status": "completed", ...}

Error Handling

All tools return errors as:

{
  "error": "descriptive error message"
}

Common errors:

"id is required" — Missing required parameter
"project not found" — Referenced resource doesn’t exist
"config must be valid JSON" — Malformed JSON parameter
"storage backend unreachable" — External system error

Getting Started

Core Concepts

Deployment

Platform Features

MCP Integration

Advanced Topics

​Projects (8 tools)

​list_projects

​get_project

​create_project

​update_project

​delete_project

​add_project_component

​remove_project_component

​clone_project

​Pipelines (6 tools)

​list_pipelines

​get_pipeline

​create_pipeline

​update_pipeline

​delete_pipeline

​execute_pipeline

​Schedules (5 tools)

​list_schedules

​get_schedule

​create_schedule

​update_schedule

​delete_schedule

​ML Models (7 tools)

​list_ml_models

​get_ml_model

​create_ml_model

​update_ml_model

​delete_ml_model

​train_ml_model

​run_inference

​recommend_model

​Digital Twins (7 tools)

​list_digital_twins

​get_digital_twin

​create_digital_twin

​update_digital_twin

​delete_digital_twin

​sync_digital_twin

​query_digital_twin

​Ontologies (6 tools)

​list_ontologies

​get_ontology

​create_ontology

​update_ontology

​delete_ontology

​generate_ontology_from_text

​extract_and_generate_ontology

​extract_from_storage

​Storage (10 tools)

​list_storage_configs

​get_storage_config

​create_storage_config

​update_storage_config

​delete_storage_config

​store_data

​retrieve_data

​update_data

​delete_data

​storage_health

​Tasks (3 tools)

​list_work_tasks

​get_work_task

​wait_for_task

​System (1 tool)

​health_check

​Common Patterns

​Chaining Operations

​Handling Async Operations

​Error Handling

Build docs developers (and LLMs) love

Projects (8 tools)

list_projects

get_project

create_project

update_project

delete_project

add_project_component

remove_project_component

clone_project

Pipelines (6 tools)

list_pipelines

get_pipeline

create_pipeline

update_pipeline

delete_pipeline

execute_pipeline

Schedules (5 tools)

list_schedules

get_schedule

create_schedule

update_schedule

delete_schedule

ML Models (7 tools)

list_ml_models

get_ml_model

create_ml_model

update_ml_model

delete_ml_model

train_ml_model

run_inference

recommend_model

Digital Twins (7 tools)

list_digital_twins

get_digital_twin

create_digital_twin

update_digital_twin

delete_digital_twin

sync_digital_twin

query_digital_twin

Ontologies (6 tools)

list_ontologies

get_ontology

create_ontology

update_ontology

delete_ontology

generate_ontology_from_text

extract_and_generate_ontology

extract_from_storage

Storage (10 tools)

list_storage_configs

get_storage_config

create_storage_config

update_storage_config

delete_storage_config

store_data

retrieve_data

update_data

delete_data

storage_health

Tasks (3 tools)

list_work_tasks

get_work_task

wait_for_task

System (1 tool)

health_check

Common Patterns

Chaining Operations

Handling Async Operations

Error Handling