Install Jentic One Beta
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the Trellis, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fruntrellis.com%2Fruntrellis" | shStep 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fruntrellis.com%2Fruntrellis" | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
What an agent can do with Trellis API.
Upload documents by URL into a project with POST /v1/assets/upload and choose a parse_strategy or chunk_strategy
Create a transform whose operations define the columns to extract, classify, or generate from each asset
Run extraction with a chosen model, trellis-scale, trellis-premium, or trellis-vertix, in document or table mode
Fetch transform results filtered by asset_ids, row_ids, and date range, with blocking reads for fresh output
GET STARTED
Retrieve or generate source references that show where a row's extracted values came from
Re-run a transform on selected assets or rows when inputs or operations change
Register webhooks and bulk event subscriptions with actions to react when processing finishes
Patterns agents use Trellis API for, with concrete tasks.
★ Document extraction agent through Jentic
An agent handling incoming paperwork needs structured fields, not raw files. Through Jentic it searches for 'extract structured data from documents', receives the Trellis operations with their input schemas, and calls POST /v1/assets/upload with the document urls and proj_id. It then calls POST /v1/transforms/{transform_id}/results to read the extracted columns for those assets and hands them to the next step of the workflow.
Call POST /v1/assets/upload with urls for three invoice files, proj_id 'proj_123', and transform_id 'tr_456', then call POST /v1/transforms/tr_456/results with those asset_ids and return the extracted totals
Building an extraction schema
A team turning loan files into a database defines what to pull once. POST /v1/projects/create makes a project with a name, and POST /v1/transforms/create takes proj_id and transform_params, where model is required and operations list the columns to extract, classify, or generate. GET /v1/transforms/{transform_id}/autoschema suggests a schema from the uploaded assets as a starting point.
Call POST /v1/projects/create with name 'Loan files', then POST /v1/transforms/create with that proj_id and transform_params model 'trellis-premium' and mode 'document', and review GET /v1/transforms/{transform_id}/autoschema
Traceable answers with source references
Reviewers want to see where each extracted value came from before approving it. POST /v1/transforms/{transform_id}/rows/{row_id}/references retrieves or generates references for a row, linking values back to the source text. PATCH /v1/transforms/{transform_id} refreshes a transform for chosen asset_ids or row_ids, with include_reference to regenerate references during the re-run.
Call POST /v1/transforms/tr_456/rows/row_789/references and list each extracted value with its referenced source passage
Event-driven processing
Pipelines that should not poll can subscribe to Trellis events. POST /v1/webhook registers a url with an auth_key that Trellis sends with each delivery, GET /v1/webhook lists registered webhooks, and POST /v1/events/subscriptions/actions/bulk creates event subscriptions with actions from an events_with_actions list.
Call POST /v1/webhook with url 'https://example.com/trellis-hook' and an auth_key, then confirm it with GET /v1/webhook
21 endpoints — the trellis api turns unstructured documents into structured tables an application can query.
METHOD
PATH
DESCRIPTION
/v1/assets/upload
Upload assets by URL
/v1/transforms/create
Create a transform
/v1/transforms/{transform_id}/results
Get transform results
/v1/transforms/{transform_id}
Refresh a transform
/v1/transforms/{transform_id}/rows/{row_id}/references
Retrieve or generate references on a row
/v1/projects/create
Create a project
/v1/webhook
Create a webhook
/v1/assets/upload
Upload assets by URL
/v1/transforms/create
Create a transform
/v1/transforms/{transform_id}/results
Get transform results
/v1/transforms/{transform_id}
Refresh a transform
/v1/transforms/{transform_id}/rows/{row_id}/references
Retrieve or generate references on a row
/v1/projects/create
Create a project
/v1/webhook
Create a webhook
What agents get from Jentic-routed access to this vendor.
Setup
Wiring Trellis by hand means putting your key in the Authorization header, pinning an API-Version header, and sequencing project, transform, upload, and results calls with the right ids. Through Jentic you install once, import Trellis from the API Directory, store the key once, and your agent calls it.
Permission scoping
Trellis puts the transform id in the URL path (/v1/transforms/{transform_id}/...), so a rule can pin your agent to one transform's results, rows, and references. You choose the operations it may call, so destructive ones like deleting a project or an asset are not included unless you add them.
Credential isolation
Your Trellis API key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
Intent-based discovery
Agents search Jentic by intent such as 'extract structured data from documents' or 'get transform results', and Jentic returns the matching Trellis operation with its input schema so the agent calls the right endpoint without reading the docs.
Alternatives and complements available in the Jentic catalogue.
Specific to using Trellis API through Jentic.
What authentication does the Trellis API use?
The Trellis API declares an apiKey scheme named APIKeyHeader in its OpenAPI spec, which sends your key in the `Authorization` header, and applies it globally. Operations also accept an optional `API-Version` header. Through Jentic the key is held by your own Jentic One instance and attached at call time.
Can I choose which model Trellis uses for extraction?
Yes. When you call POST /v1/transforms/create, transform_params requires a model, which the spec says must be one of trellis-scale, trellis-premium, or trellis-vertix, and a mode of document or table. The operations list defines the columns, and at least one parse column of the assets type is required.
What are the rate limits for the Trellis API?
The OpenAPI spec does not specify rate limits. Check the Trellis site at https://runtrellis.com for current limits before uploading large batches of assets.
How do I extract data from documents through Jentic?
Search Jentic for 'extract structured data from documents', load POST /v1/assets/upload, and execute it with the document urls, your proj_id, and transform_id. Then call POST /v1/transforms/{transform_id}/results with the asset_ids to read the extracted rows. To run it on your own infrastructure, install Jentic One from its GitHub repo.
Is there a Trellis MCP server?
You don't need an MCP server to give your agent Trellis. Jentic connects it directly from the API Directory: import it, store your API key once, and your agent finds the upload, transform, and results operations it needs on demand, without loading extra tool definitions into its context.
Can I limit what my agent is allowed to do with the Trellis API?
Yes. For an extraction worker, write a rule that allows POST /v1/assets/upload and POST /v1/transforms/{transform_id}/results under one transform id. The agent can submit documents and read that transform's output but cannot create transforms, delete projects, or register webhooks, and each call it makes is logged.
For Agents
Upload documents into a Trellis project, run a transform that extracts, classifies, and generates column values, and read structured results with source references.
Use for: Extract invoice numbers, totals, and due dates from a batch of uploaded invoices, I need to upload contracts from S3 URLs into a Trellis project, Create a transform that classifies documents by type, Get the results of my Trellis transform as structured rows
Not supported: Does not store documents long term, e-sign contracts, or run OCR-only scanning hardware. Use for extracting and classifying structured data from uploaded documents only.
The Trellis API turns unstructured documents into structured tables an application can query. You upload assets by URL into a project, define a transform whose operations are columns that extract, classify, or generate values, and pick a model such as trellis-scale, trellis-premium, or trellis-vertix in document or table mode. Results come back per asset and row with optional references to the source text, and webhooks plus event subscriptions notify your system as processing completes.
This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.
Base layer of spec validity and structural soundness.
Aggregated quality score from linter diagnostics, weighted by severity.
Percentage of `$ref` references that resolve successfully.
Checks whether the API description parses successfully and conforms to its declared specification (e.g., OpenAPI).
Structural correctness score based on schema issues using logarithmic dampening.
Clarity, completeness, and ingestion readiness for developers and tooling.
How richly the API is illustrated with examples.
Percentage of examples that conform to their schemas.
Percentage of operations with complete response definitions (success, client error, server error).
Health of API ingestion, bundling, and resolution within Jentic pipelines.
Semantic breadth, depth, and agent comprehension for AI systems.
Coverage of descriptions across API elements.
Coverage of RFC 9457 Problem Details for error responses.
Coverage, uniqueness, and casing consistency of operationIds for AI inference.
Coverage of summaries across operations/tags/info.
Functional utility, complexity comfort, and AI orchestration readiness.
Agent comfort level based on API operational and structural complexity.
Trust, risk posture, and security compliance.
Average quality of security schemes based on authentication method strength (weakest link for OAuth2).
Findability, semantic richness, and reasoning readiness.
Clarity and depth of descriptions across API elements.
Score it yourself
Every API in the directory is allowlisted, so you can re-score it with no key required.
npx @jentic/api-scorecard-cli score <openapi-url>