V2 Search

POST https://api.codefundi.app/v2/search

There is no separate /v2/research route. Research is this endpoint with chat: true.

Mode	`chat`	Content-Type	History category
Search	`false` (default)	`application/json`	`search`
Research	`true`	`application/x-ndjson`	`research`

Credits: 5 credits per search. Research adds model usage on top. See Credits and tiers.

Product: Search, Research.

Request body

Field	Required	Type	Description
`query`	yes	string	Search text (2–1000 characters).
`scope`	no	enum	`all` (default), `repos`, `files`, `code`, `functions`.
`scan_mode`	no	enum	`semantic` (default), `grep_docs`, `grep_code`.
`repo_ids`	no	uuid[]	Restrict to indexed repository IDs.
`repo_urls`	no	string[]	Clone URLs; merged with `repo_ids`, max 25 combined.
`filters`	no	object	`file_types`, `file_paths`, `dependencies`, `function_names`, `has_functions`, `min_lines`, `max_lines`, `visibility`.
`similarity_threshold`	no	number	0–1 optional floor for semantic matches.
`fields`	no	enum	Documentation depth on hits: `basic`, `summary`, or `full`.
`chat`	no	boolean	Default `false`. `true` enables NDJSON research.
`model`	no	string	Model id when `chat` is true.

Responses: 200; 400, 401, 402, 500.

Search example (`chat: false`)

cURL

search.sh


curl -sS -X POST "https://api.codefundi.app/v2/search" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d "{\"query\":\"authentication middleware\",\"scope\":\"all\",\"scan_mode\":\"semantic\",\"chat\":false,\"fields\":\"summary\"}"

JavaScript


interface SearchRequest {
  query: string
  scope?: 'all' | 'repos' | 'files' | 'code' | 'functions'
  scan_mode?: 'semantic' | 'grep_docs' | 'grep_code'
  repo_ids?: string[]
  repo_urls?: string[]
  chat?: boolean
  fields?: 'basic' | 'summary' | 'full'
  model?: string
}

const body: SearchRequest = {
  query: 'authentication middleware',
  scope: 'all',
  scan_mode: 'semantic',
  chat: false,
  fields: 'summary'
}

const res = await fetch('https://api.codefundi.app/v2/search', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'X-API-Key': 'YOUR_API_KEY'
  },
  body: JSON.stringify(body)
})
//    ^?

Python

search.py


import json
import urllib.request
 
payload = {
    "query": "authentication middleware",
    "scope": "all",
    "scan_mode": "semantic",
    "chat": False,
    "fields": "summary",
}
req = urllib.request.Request(
    "https://api.codefundi.app/v2/search",
    data=json.dumps(payload).encode(),
    headers={"Content-Type": "application/json", "X-API-Key": "YOUR_API_KEY"},
    method="POST",
)
with urllib.request.urlopen(req) as resp:
    print(resp.read().decode())

Research example (`chat: true`)

NDJSON lines: objects of type search (results, total, meta), then chunk (text), then done (model, context_files) or error (message).

research.sh


curl -sS -N -X POST "https://api.codefundi.app/v2/search" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Accept: application/x-ndjson" \
  -d "{\"query\":\"How is auth enforced?\",\"scope\":\"all\",\"scan_mode\":\"semantic\",\"chat\":true,\"model\":\"gpt-4o-mini\"}"

For chat: true, send Accept: application/x-ndjson, read line-by-line, and parse each JSON object (type discriminates rows).