Create retrieve

POST /v1/knowledge/files/retrieve

Knowledge → · All API references → · Get started →

Address	`https://api.hanzo.ai/v1/knowledge/files/retrieve`
Method	`POST`
Operation	`post_knowledge_files_retrieve`
Auth	`Authorization: Bearer $HANZO_API_KEY`

Grounds an answer in the caller's org files, table of contents first: candidate documents are found by searching their passages; a model reads their tables of contents — titles and one-line summaries — and picks the sections the answer is in; hybrid search drills into the passages of those sections; and the graph expands to the sections they link to or name the same entities as, in any file of the workspace. Every passage cites its file › section › paragraph, so an answer can say exactly where it came from. This is what a chat runs before it answers about an attached file, and what an agent runs before it answers about a workspace.

Request

4 fields, body application/json (required).

Field	In	Type	Required	Description
`files`	body	string[]	—	Files are the files the question is about — the files a chat message references.
`limit`	body	integer (int64)	—	Limit bounds the passages drilled out of the chosen sections.
`project`	body	string	—	Project restricts retrieval to files indexed under one project scope.
`query`	body	string	—	Query is the question to answer.

Response

Status	Body	Meaning
`200`	knowledge.retrieveOut	ok
`default`	problem-details	refused

200 body — 31 fields.

Field	In	Type	Always	Description
`degraded`	body	boolean	—	Degraded is true when a search leg failed along the way.
`passages`	body	knowledge.citedPassage[]	—	Passages are the passages drilled out of those sections, then those the graph reached, each citing file › section › paragraph.
`passages[].cite`	body	string	—	Cite is how an answer cites it: file › section path › ¶part.
`passages[].file`	body	knowledge.fileBrief	—
`passages[].file.bucket`	body	string	—	Bucket and Key are where its bytes are: GET /v1/s3/buckets/{bucket}/objects/{key} answers a signed download URL.
`passages[].file.id`	body	string	—	ID is the file's id.
`passages[].file.key`	body	string	—	Key is the object's key in Bucket.
`passages[].file.name`	body	string	—	Name is the file's name.
`passages[].file.type`	body	string	—	Type is the file's media type.
`passages[].part`	body	integer (int64)	—	Part is the passage's place in its section, from 1.
`passages[].score`	body	number (double)	—	Score is the fused rank score, comparable within one response only.
`passages[].section`	body	knowledge.sectionBrief	—
`passages[].section.id`	body	integer (int64)	—	ID is the section's number in its file's table of contents; 0 is the document itself.
`passages[].section.path`	body	string	—	Path is the section's place in its document: the titles from the document's own down to it, joined by " › ".
`passages[].section.title`	body	string	—	Title is the section's heading.
`passages[].text`	body	string	—	Text is the passage: a span of its section of about 2000 bytes, never crossing into another section.
`passages[].via`	body	string	—	Via is how retrieval reached it: search (it matched), toc (it is in a section the table of contents pointed at) or graph (a linked section).
`passages[].why`	body	string	—	Why says, for a passage reached through the graph, which edge led there.
`picked`	body	string	—	Picked says how the sections were chosen: whole when the documents were short enough to read every section, model when a model read the tables of contents, search when the sections of the best passages stood in — the documents had no headings to choose by, or no model answered.
`sections`	body	knowledge.pickedSection[]	—	Sections are the sections chosen from the candidate documents' tables of contents.
`sections[].file`	body	knowledge.fileBrief	—
`sections[].file.bucket`	body	string	—	Bucket and Key are where its bytes are: GET /v1/s3/buckets/{bucket}/objects/{key} answers a signed download URL.
`sections[].file.id`	body	string	—	ID is the file's id.
`sections[].file.key`	body	string	—	Key is the object's key in Bucket.
`sections[].file.name`	body	string	—	Name is the file's name.
`sections[].file.type`	body	string	—	Type is the file's media type.
`sections[].section`	body	knowledge.sectionBrief	—
`sections[].section.id`	body	integer (int64)	—	ID is the section's number in its file's table of contents; 0 is the document itself.
`sections[].section.path`	body	string	—	Path is the section's place in its document: the titles from the document's own down to it, joined by " › ".
`sections[].section.title`	body	string	—	Title is the section's heading.
`sections[].summary`	body	string	—	Summary is its one-line account.

Failure carries the platform error shape — see Errors.

Examples

CLI

SDK

HTTP

MCP

hanzo has no subcommand for this operation — the CLI serves only what cloud's live route table confirms. Use HTTP or an SDK.

.ts

.py

.go

.rs

.java

import { Configuration, KnowledgeApi } from 'hanzoai';

const api = new KnowledgeApi(new Configuration({ accessToken: process.env.HANZO_API_KEY }));
const { data } = await api.postKnowledgeFilesRetrieve({ files: ["<files>"], limit: 0 });

from hanzoai.cloud import ApiClient, Configuration
from hanzoai.cloud.api import KnowledgeApi

client = ApiClient(Configuration(access_token=os.environ["HANZO_API_KEY"]))
result = KnowledgeApi(client).post_knowledge_files_retrieve(files=["<files>"], limit=0)

cfg := hanzoai.NewConfiguration()
cfg.AddDefaultHeader("Authorization", "Bearer "+os.Getenv("HANZO_API_KEY"))
client := hanzoai.NewAPIClient(cfg)

resp, _, err := client.KnowledgeAPI.PostKnowledgeFilesRetrieve(context.Background()).Execute()
if err != nil {
	return err
}

use hanzo_client::apis::{configuration::Configuration, knowledge_api};

let mut cfg = Configuration::new();
cfg.bearer_access_token = std::env::var("HANZO_API_KEY").ok();

let result = knowledge_api::post_knowledge_files_retrieve(&cfg, Default::default()).await?;

import ai.hanzo.cloud.ApiClient;
import ai.hanzo.cloud.api.KnowledgeApi;

ApiClient client = new ApiClient();
client.setBearerToken(System.getenv("HANZO_API_KEY"));

var result = new KnowledgeApi(client).postKnowledgeFilesRetrieve();

curl -X POST https://api.hanzo.ai/v1/knowledge/files/retrieve \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
       "files": [
         "<files>"
       ],
       "limit": 0
     }'

MCP reaches knowledge through the knowledge tool, which names its 9 operations with its own verbs — this one among them, under a name only MCP declares. describe explains any of them:

curl -X POST https://api.hanzo.ai/v1/mcp \
  -H "Content-Type: application/json" \
  -d '{
       "jsonrpc": "2.0",
       "id": 1,
       "method": "tools/call",
       "params": {
         "name": "describe",
         "arguments": {
           "op": "list_knowledge_connectors"
         }
       }
     }'

Knowledge API · All Hanzo APIs · Interactive reference

Was this page useful?

PreviousSync connectors NextSearch files