Loaders

snowloader provides six loaders, each targeting a specific ServiceNow table. All loaders share the same interface inherited from BaseSnowLoader:

• load() - returns a list of all matching documents

• lazy_load() - yields documents one at a time (memory efficient)

• load_since(datetime) - delta sync, fetches only updated records

IncidentLoader

Loads IT incidents from the incident table. Documents include the incident number, summary, description, state, priority, category, assignment details, and relevant timestamps.

from snowloader import SnowConnection, IncidentLoader

conn = SnowConnection(instance_url="...", username="...", password="...")

loader = IncidentLoader(
    connection=conn,
    query="active=true^priority<=2",
    include_journals=True,  # Append work notes and comments
)

for doc in loader.lazy_load():
    print(doc.metadata["number"], doc.metadata["state"])

With journal entries:

When include_journals=True, work notes and comments from sys_journal_field are appended to the document text. This is useful for capturing the full investigation history.

KnowledgeBaseLoader

Loads articles from the kb_knowledge table. HTML content is automatically stripped using the built-in cleaner (no BeautifulSoup dependency). Falls back to the wiki field when text is empty.

from snowloader import KnowledgeBaseLoader

loader = KnowledgeBaseLoader(
    connection=conn,
    query="workflow_state=published",
)

for doc in loader.lazy_load():
    # Clean plain text, no HTML tags
    print(doc.page_content[:300])

CMDBLoader

Loads Configuration Items from any CMDB class table. The most powerful loader - it can optionally traverse the relationship graph to show how CIs depend on each other.

from snowloader import CMDBLoader

# Load servers with their dependency graph
loader = CMDBLoader(
    connection=conn,
    ci_class="cmdb_ci_server",
    include_relationships=True,
    max_relationship_workers=2,  # Concurrent threads for relationship queries
)

for doc in loader.lazy_load():
    print(doc.page_content)
    # Document text includes:
    #   -> db-prod-01 (Depends on::Used by)
    #   <- load-balancer-01 (Depends on::Used by)

    # Structured data also in metadata
    for rel in doc.metadata.get("relationships", []):
        print(f"  {rel['direction']}: {rel['target']} ({rel['type']})")

Parameters:

• ci_class - CMDB class table (default: "cmdb_ci"). Use "cmdb_ci_server", "cmdb_ci_service", etc. for specific classes.

• include_relationships - when True, fetches outbound and inbound relationships from cmdb_rel_ci. Adds 2 API calls per CI.

• max_relationship_workers - number of concurrent threads for relationship queries (default: 2).

ChangeLoader

Loads change requests from the change_request table. Documents emphasize the change type, risk level, implementation window, and assignment.

from snowloader import ChangeLoader

loader = ChangeLoader(connection=conn, query="state=2")  # Scheduled
docs = loader.load()

ProblemLoader

Loads problem records from the problem table. Documents highlight root cause, known error status, and fix notes - the fields most valuable for LLM-powered incident correlation.

from snowloader import ProblemLoader

loader = ProblemLoader(connection=conn, query="known_error=true")
for doc in loader.lazy_load():
    if doc.metadata["known_error"]:  # Python bool, not string
        print(f"{doc.metadata['number']}: {doc.page_content[:200]}")

CatalogLoader

Loads service catalog items from the sc_cat_item table. Useful for building LLM-powered service desk chatbots that help users find and request services.

from snowloader import CatalogLoader

loader = CatalogLoader(connection=conn, query="active=true")
docs = loader.load()