The LlamaParse connector lets your AI agent act in each user's LlamaParse account. Each user connects their own LlamaParse access token once, and Scalekit sends it with every call, so your agent never handles credentials. It comes with 22 tools.
- Tools
- 22
- What they doRead · write · destructive
- 12 · 7 · 312 read7 write3 destructive
- Users sign in with
- Access token
Setup
Install the SDK
Terminal window npm install @scalekit-sdk/node dotenvTerminal window pip install scalekit-sdk-python python-dotenvSet your credentials
Add your Scalekit credentials to your
.envfile. Find values in app.scalekit.com > Developers > API Credentials..env SCALEKIT_ENVIRONMENT_URL=<your-environment-url>SCALEKIT_CLIENT_ID=<your-client-id>SCALEKIT_CLIENT_SECRET=<your-client-secret>Create the LlamaParse connection
In AgentKit > Connections, create a LlamaParse connection. The name you give it is the
connection_nameyour code passes. See Configure connections.Authorize a user and make your first call
quickstart.mts import { ScalekitClient } from '@scalekit-sdk/node'import 'dotenv/config'import { createInterface } from 'node:readline/promises'const scalekit = new ScalekitClient(process.env.SCALEKIT_ENVIRONMENT_URL,process.env.SCALEKIT_CLIENT_ID,process.env.SCALEKIT_CLIENT_SECRET,)const actions = scalekit.actionsconst connector = 'llamaparse'const identifier = 'user_123'// Generate an authorization link for the userconst { link } = await actions.getAuthorizationLink({ connectionName: connector, identifier })console.log('Authorize LlamaParse:', link)const rl = createInterface({ input: process.stdin, output: process.stdout })await rl.question('Press Enter after authorizing...')rl.close()// Make your first callconst result = await actions.executeTool({connector,identifier,toolName: 'llamaparse_classify_jobs_list',toolInput: {},})console.log(result)Terminal window npx tsx quickstart.mtsquickstart.py import osfrom scalekit import ScalekitClientfrom dotenv import load_dotenvload_dotenv()scalekit_client = ScalekitClient(env_url=os.getenv("SCALEKIT_ENVIRONMENT_URL"),client_id=os.getenv("SCALEKIT_CLIENT_ID"),client_secret=os.getenv("SCALEKIT_CLIENT_SECRET"),)actions = scalekit_client.actionsconnection_name = "llamaparse"identifier = "user_123"# Generate an authorization link for the userlink_response = actions.get_authorization_link(connection_name=connection_name,identifier=identifier,)print("Authorize LlamaParse:", link_response.link)input("Press Enter after authorizing...")# Make your first callresult = actions.execute_tool(tool_input={},tool_name="llamaparse_classify_jobs_list",connection_name=connection_name,identifier=identifier,)print(result)Terminal window python quickstart.pyEach user opens the link once and enters their LlamaParse credentials there. See Authorize a user for the full flow and statuses.
Tools
Pass the exact name toexecute_toolllamaparse_classify_job_getGet a classify job's status and, when finished, the chosen document type with its confidence and reasoning.Read-onlyGet Classify Job
Get a classify job's status and, when finished, the chosen document type with its confidence and reasoning. Returns id, status, error_message and the classification result. Use this to poll a classification and read the result.
Inputs
job_idstringrequired- Classify job ID returned by llamaparse_classify_job_create or listed by llamaparse_classify_jobs_list. Example: clj-5b8e2f9c3a1d
expandarray- Extra fields to include: usage. Example: ["usage"]
llamaparse_classify_jobs_listList classify jobs in the project, optionally filtered by status and creation time.Read-onlyList Classify Jobs
List classify jobs in the project, optionally filtered by status and creation time. Returns items (each with id, status and created_at) and next_page_token. Use this to find earlier classifications. Use llamaparse_classify_job_get for a result.
Inputs
created_at_on_or_afterstring- Only jobs created at or after this time (ISO 8601). Example: 2026-11-01T00:00:00Z
created_at_on_or_beforestring- Only jobs created at or before this time (ISO 8601). Example: 2026-11-15T23:59:59Z
page_sizeinteger- Jobs per page, 1 to 100. Example: 20
page_tokenstring- Token from next_page_token of the previous response, to get the next page. Example: eyJvZmZzZXQiOjIwfQ
statusstring- Only return classify jobs in this status: PENDING, RUNNING, COMPLETED, FAILED, CANCELLED. Example: COMPLETEDone of
PENDINGRUNNINGCOMPLETEDFAILEDCANCELLED
llamaparse_extract_job_getGet an extract job's status and, when finished, the extracted data.Read-onlyGet Extract Job
Get an extract job's status and, when finished, the extracted data. Returns id, status, error_message and extract_result (data matching the schema), plus any expanded fields. Use this to poll an extraction and read the result.
Inputs
job_idstringrequired- Extract job ID returned by llamaparse_extract_job_create or listed by llamaparse_extract_jobs_list. Example: exj-7f2c9d1a4b6e
expandarray- Extra fields to include: configuration, extract_metadata (citations and confidence scores), usage. Example: ["extract_metadata"]
llamaparse_extract_jobs_listList extract jobs in the project, optionally filtered by status, source document or creation time.Read-onlyList Extract Jobs
List extract jobs in the project, optionally filtered by status, source document or creation time. Returns items (each with id, status, file_input and created_at) and next_page_token. Use this to find earlier extractions. Use llamaparse_extract_job_get for a job's result.
Inputs
created_at_on_or_afterstring- Only jobs created at or after this time (ISO 8601). Example: 2026-11-01T00:00:00Z
created_at_on_or_beforestring- Only jobs created at or before this time (ISO 8601). Example: 2026-11-15T23:59:59Z
file_inputstring- Only jobs run on this parse job ID or file ID. Example: pjb-3k9x2m7q1v8c
page_sizeinteger- Jobs per page, 1 to 1000. Example: 20
page_tokenstring- Token from next_page_token of the previous response, to get the next page. Example: eyJvZmZzZXQiOjIwfQ
statusstring- Only return extract jobs in this status: PENDING, THROTTLED, RUNNING, COMPLETED, FAILED, CANCELLED. Example: COMPLETEDone of
PENDINGTHROTTLEDRUNNINGCOMPLETEDFAILEDCANCELLED
llamaparse_extract_schema_validateCheck whether a JSON Schema can be used for extraction.Read-onlyValidate Extraction Schema
Check whether a JSON Schema can be used for extraction. Returns the validated (normalized) schema, or a 422 error describing what is wrong. Use this before llamaparse_extract_job_create.
Inputs
data_schemaobjectrequired- JSON Schema to check for use with extract jobs. Example: {"type": "object", "properties": {"total": {"type": "number"}}}
llamaparse_organizations_listList the LlamaCloud organizations the API key's user belongs to.Read-onlyList Organizations
List the LlamaCloud organizations the API key's user belongs to. Returns items with id, name and created_at, plus next_page_token. Use this with llamaparse_projects_list to see account structure.
Inputs
namestring- Only organizations with this name. Example: Acme Corp
page_sizeinteger- Organizations per page. Example: 20
page_tokenstring- Token from next_page_token of the previous response. Example: eyJvZmZzZXQiOjIwfQ
llamaparse_parse_job_getGet a parse job's status and, with expand, its parsed output as markdown, text, structured items or metadata.Read-onlyGet Parse Job
Get a parse job's status and, with expand, its parsed output as markdown, text, structured items or metadata. Returns job (id, status, error_message) plus the requested outputs, for example markdown.pages[].markdown. Use this to poll a job and read the result. Use llamaparse_parse_jobs_list to find job IDs.
Inputs
job_idstringrequired- Parse job ID returned by llamaparse_parse_job_create or listed by llamaparse_parse_jobs_list. Example: pjb-3k9x2m7q1v8c
expandarray- Output to include with the job: text, markdown, items, metadata, forms, job_metadata, usage, or one of the *_content_metadata fields (which carry presigned download URLs, for example images_content_metadata or xlsx_content_metadata). Without expand only the job status is returned. Example: ["markdown", "usage"]
image_filenamesstring- Comma-separated image filenames to include when expanding images_content_metadata. Example: image_0.png,image_1.jpg
llamaparse_parse_jobs_listList parse jobs in the project, newest first, optionally filtered by status and creation time.Read-onlyList Parse Jobs
List parse jobs in the project, newest first, optionally filtered by status and creation time. Returns items (each with id, status, tier, created_at and the source) and next_page_token. Use this to find an earlier job. Use llamaparse_parse_job_get for a job's output.
Inputs
created_at_on_or_afterstring- Only jobs created at or after this time (ISO 8601). Example: 2026-11-01T00:00:00Z
created_at_on_or_beforestring- Only jobs created at or before this time (ISO 8601). Example: 2026-11-15T23:59:59Z
page_sizeinteger- Jobs per page, 1 to 1000. Example: 20
page_tokenstring- Token from next_page_token of the previous response, to get the next page. Example: eyJvZmZzZXQiOjIwfQ
statusstring- Only return parse jobs in this status: PENDING, RUNNING, COMPLETED, FAILED, CANCELLED. Example: COMPLETEDone of
PENDINGRUNNINGCOMPLETEDFAILEDCANCELLED
llamaparse_parse_versions_listList the available versions for each parse tier.Read-onlyList Parse Versions
List the available versions for each parse tier. Returns each tier with its dated versions and which one latest points to. Use this to pin a version in llamaparse_parse_job_create.
Inputs
This tool takes no inputs.
llamaparse_project_getGet one LlamaCloud project.Read-onlyGet Project
Get one LlamaCloud project. Returns id, name, organization_id and timestamps. Use llamaparse_projects_list to find the ID.
Inputs
project_idstringrequired- Project ID (UUID) from llamaparse_projects_list. Example: 7a8b9c0d-1e2f-4a3b-8c4d-5e6f7a8b9c0d
llamaparse_projects_listList the LlamaCloud projects the API key can access.Read-onlyList Projects
List the LlamaCloud projects the API key can access. Returns items with id, name, organization_id and created_at, plus next_page_token. Use this to see where jobs run. API keys are scoped to a project, so this usually returns one.
Inputs
namestring- Only projects with this name. Example: Default
organization_idstring- Only projects in this organization. Example: 1d2c3b4a-5e6f-4a7b-8c9d-0e1f2a3b4c5d
page_sizeinteger- Projects per page. Example: 20
page_tokenstring- Token from next_page_token of the previous response. Example: eyJvZmZzZXQiOjIwfQ
llamaparse_supported_extensions_listList the file extensions LlamaParse can parse.Read-onlyList Supported File Types
List the file extensions LlamaParse can parse. Returns an array of supported extensions with their content types. Use this to check a document type before llamaparse_parse_job_create.
Inputs
This tool takes no inputs.
llamaparse_classify_job_cancelCancel a pending or running classify job.WriteCancel Classify Job
Cancel a pending or running classify job. Returns the job with its updated status. Use this to stop a classification you no longer need.
- Idempotent
Inputs
job_idstringrequired- Classify job ID returned by llamaparse_classify_job_create or listed by llamaparse_classify_jobs_list. Example: clj-5b8e2f9c3a1d
llamaparse_classify_job_createStart a LlamaClassify job that labels a parsed document or file with one of the document types you describe, spending credits.WriteClassify Document
Start a LlamaClassify job that labels a parsed document or file with one of the document types you describe, spending credits. Returns the job with id and status. Poll llamaparse_classify_job_get until COMPLETED to read the chosen type and reasoning. Use this to route documents by type. Use llamaparse_extract_job_create to pull fields out of them.
Inputs
file_inputstringrequired- Parse job ID (from llamaparse_parse_job_create) or LlamaCloud file ID of the document to classify. To work from a URL, parse it first and pass the parse job ID. Example: pjb-3k9x2m7q1v8c
configuration_idstring- ID of a saved classify configuration to use instead of rules. Example: cfg-6d3a8b2e9f
rulesarray- Document types to choose from, each with a type label (up to 50 characters) and a description (up to 2000) of what that type looks like. Required unless configuration_id is set. Example: [{"type": "invoice", "description": "A bill with line items, amounts and a total due"}, {"type": "contract", "description": "A legal agreement with parties, terms and signatures"}]
transaction_idstring- Idempotency key, up to 255 characters, unique in the project. Reusing it returns the original job. Example: classify-inbox-2026-11-15-001
llamaparse_extract_job_cancelCancel a pending or running extract job.WriteCancel Extract Job
Cancel a pending or running extract job. Returns the job with its updated status. Use this to stop an extraction you no longer need.
- Idempotent
Inputs
job_idstringrequired- Extract job ID returned by llamaparse_extract_job_create or listed by llamaparse_extract_jobs_list. Example: exj-7f2c9d1a4b6e
llamaparse_extract_job_createStart a LlamaExtract job that pulls structured data matching a JSON Schema out of a parsed document or file, spending credits.WriteExtract Structured Data
Start a LlamaExtract job that pulls structured data matching a JSON Schema out of a parsed document or file, spending credits. Returns the job with id and status. Poll llamaparse_extract_job_get until COMPLETED to read extract_result. Use this after llamaparse_parse_job_create to get fields such as totals, dates or names. Use llamaparse_classify_job_create to label the document type instead.
Inputs
file_inputstringrequired- Parse job ID (from llamaparse_parse_job_create) or LlamaCloud file ID of the document to extract from. To work from a URL, parse it first and pass the parse job ID. Example: pjb-3k9x2m7q1v8c
cite_sourcesboolean- If true, return citations for each value under extract_metadata. Example: true
confidence_scoresboolean- If true, return confidence scores under extract_metadata. Example: true
configuration_idstring- ID of a saved extract configuration to use instead of data_schema and the other settings. Example: cfg-8c1e4f7a2b
data_schemaobject- JSON Schema of the fields to extract. Required unless configuration_id is set. Check it with llamaparse_extract_schema_validate or draft one with llamaparse_extract_schema_generate. Example: {"type": "object", "properties": {"invoice_number": {"type": "string"}, "total": {"type": "number"}}}
max_pagesinteger- Maximum number of pages to process, starting from page 1. target_pages wins if both are set. Example: 10
system_promptstring- Extra instructions that guide the extraction. Example: Amounts are in EUR; use the grand total including VAT.
target_pagesstring- Pages to process, 1-based, as a comma-separated list of pages and ranges. Omit for all pages. Example: 1,3,5-8
tierstring- Extract tier: cost_effective (5 credits/page), agentic (15), agentic_plus (50) or turbo (35). Example: cost_effectiveone of
cost_effectiveagenticagentic_plusturbo
llamaparse_extract_schema_generateDraft an extraction JSON Schema from a plain-language description, a sample file or an existing schema.WriteGenerate Extraction Schema
Draft an extraction JSON Schema from a plain-language description, a sample file or an existing schema. Returns the generated configuration with its data_schema. Use this before llamaparse_extract_job_create when you do not have a schema yet. Use llamaparse_extract_schema_validate to check a schema you wrote.
Inputs
data_schemaobject- Existing schema to refine or extend. Example: {"type": "object", "properties": {"invoice_number": {"type": "string"}}}
file_idstring- LlamaCloud file ID of a sample document to base the schema on. Example: 4f6b2c1e-9a7d-4e3b-8c5f-1a2b3c4d5e6f
namestring- Name for the generated configuration, up to 255 characters. Auto-generated if empty. Example: Invoice fields
promptstring- Plain-language description of the data to extract. Example: Invoice number, issue date, vendor name, line items with description and amount, and the total
llamaparse_parse_job_cancelCancel a pending or running parse job.WriteCancel Parse Job
Cancel a pending or running parse job. Returns the job with its updated status. Use this to stop a job you no longer need. Use llamaparse_parse_job_delete to remove a finished job.
- Idempotent
Inputs
job_idstringrequired- Parse job ID returned by llamaparse_parse_job_create or listed by llamaparse_parse_jobs_list. Example: pjb-3k9x2m7q1v8c
llamaparse_parse_job_createStart a LlamaParse job that downloads a document from a public URL and parses it into markdown, text and structured items, spending credits.WriteParse Document from URL
Start a LlamaParse job that downloads a document from a public URL and parses it into markdown, text and structured items, spending credits. Returns the job with id and status. Parsing runs asynchronously; poll llamaparse_parse_job_get until status is COMPLETED and request the output with expand. Use this first for any document. Use llamaparse_extract_job_create or llamaparse_classify_job_create on the resulting job ID for structured data or document types.
Inputs
source_urlstringrequired- Public URL of the document to parse (PDF, Word, PowerPoint, spreadsheet, image, HTML and more; see llamaparse_supported_extensions_list). LlamaParse downloads it, so it must be reachable without login. Example: https://example.com/reports/annual-report-2025.pdf
tierstringrequired- Parsing tier: fast (rule-based, cheapest, text only), cost_effective (balanced), agentic (AI-powered, supports custom_prompt) or agentic_plus (highest accuracy). Pass configured with configuration_id to use the saved configuration's tier. Example: cost_effectiveone of
fastcost_effectiveagenticagentic_plusconfigured configuration_idstring- ID of a saved parse configuration. Its settings apply, and tier and version default to the saved values (pass configured for both). Example: cfg-2a7d9e4b1c
custom_promptstring- Instructions for the AI tiers (cost_effective, agentic, agentic_plus) on how to parse, for example how to transcribe tables or what to ignore. Not used by the fast tier. Example: Output every table as markdown and skip page headers and footers.
disable_cacheboolean- If true, ignore cached results and parse again. Use when the document at the URL changed. Example: false
max_pagesinteger- Maximum number of pages to process, starting from page 1. target_pages wins if both are set. Example: 10
output_optionsobject- Output settings, for example markdown table handling, spatial text layout, or images to save. Example: {"markdown": {"tables": {"output_tables_as_markdown": true}, "annotate_links": true}}
processing_optionsobject- Processing settings, for example OCR languages or text to ignore. Example: {"ocr_parameters": {"languages": ["en", "de"]}, "ignore": {"ignore_hidden_text": true}}
target_pagesstring- Pages to process, 1-based, as a comma-separated list of pages and ranges. Omit for all pages. Example: 1,3,5-8
versionstring- Tier version: latest, or a dated version from llamaparse_parse_versions_list (for example 2026-11-09). Pass configured with configuration_id. Default: latest. Example: latestdefault
latest
llamaparse_classify_job_deleteDelete a classify job and its result.DestructiveDelete Classify Job
Delete a classify job and its result. Returns a confirmation. Use this to remove classification data.
- Idempotent
Inputs
job_idstringrequired- Classify job ID returned by llamaparse_classify_job_create or listed by llamaparse_classify_jobs_list. Example: clj-5b8e2f9c3a1d
llamaparse_extract_job_deleteDelete an extract job and its results.DestructiveDelete Extract Job
Delete an extract job and its results. Returns a confirmation. Use this to remove extracted data.
- Idempotent
Inputs
job_idstringrequired- Extract job ID returned by llamaparse_extract_job_create or listed by llamaparse_extract_jobs_list. Example: exj-7f2c9d1a4b6e
forceboolean- If true, delete even if the job is still running. Default: false. Example: false
llamaparse_parse_job_deleteDelete a parse job and its stored results.DestructiveDelete Parse Job
Delete a parse job and its stored results. Returns a confirmation; 409 if the job is still running. Use this to remove parsed data. Use llamaparse_parse_job_cancel to stop a running job first.
- Idempotent
Inputs
job_idstringrequired- Parse job ID returned by llamaparse_parse_job_create or listed by llamaparse_parse_jobs_list. Example: pjb-3k9x2m7q1v8c
No tools match.