Skip to main content
GET
/
api
/
credits
/
config
Read the current credit configuration
curl --request GET \
  --url https://app.puffle.ai/api/credits/config \
  --header 'Authorization: Bearer <token>'
{
  "costs": {
    "FIND_PERSON": 1,
    "IMPORT_LEAD": 1,
    "EMAIL_LOOKUP": 2,
    "PHONE_LOOKUP": 8,
    "LINKEDIN_PROFILE": 1,
    "LINKEDIN_POSTS": 1,
    "COMPANY_NEWS": 4,
    "PERSON_NEWS": 4,
    "HIRING_STATUS": 1,
    "WEBSITE_ENRICHMENT": 1,
    "SITE_CRAWL_ENRICHMENT": 4,
    "CUSTOM_ENRICHMENT_AI": 1,
    "CUSTOM_ENRICHMENT_WEB": 1,
    "DEEP_RESEARCH": 40,
    "AI_MESSAGE_GENERATION": 2,
    "AGENT_SEARCH": 10,
    "SEARCH_CONTINUE": 5
  }
}

​
Overview

Returns the full map of action → credits-consumed-per-unit. This is a pure static read with no side effects. Intended for UIs and agents that want to surface cost estimates before calling a billable endpoint. One credit approximately equals $0.05 face value. Multiply costs[action] * count to get the total for a batch operation (e.g. FIND_PERSON * 50 for a 50-result lead search, DEEP_RESEARCH * 10 for 10 research reports).

​
The credit system

Puffle charges credits for actions that call paid third-party services (enrichment, deep research, signal scans, LLM-powered generation) or consume meaningful compute. Plans (free / pro / enterprise) top up the balance; actions deduct from it. When the balance hits zero, billable endpoints return 402. Agents should pre-check balance before calling any billable endpoint. The pattern is:
  1. Call this endpoint once at session start (or cache it) to know the per-action price.
  2. Before each billable operation, call createCreditPreview with { action, count } to confirm affordability against the live balance.
  3. If the preview reports sufficient: false, stop and escalate to the human with the exact shortfall — never partially-execute.
See the credit system in the Agent Playbook for the full pattern.

​
AI agent notes

Prices are stable but not immutable. The list updates as we add new integrations or retune pricing; cache for the session, not forever.Use preview, not config, for affordability. This endpoint tells you the price. createCreditPreview tells you whether the caller can afford it against their current balance — that is the call an agent actually needs before proceeding with a billable action.New free workspaces need provisioning. If a fresh workspace returns 402 on billable endpoints even though the balance should be 1,000, fall back to createCreditProvision to attach the free plan before retrying.

Authorizations

​
Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Response

The full cost table. Keys are action identifiers, values are positive integers (credits per unit of work).

​
costs
object
required

Map of action → credits consumed per unit. Multiply by batch size for a per-operation estimate (e.g. FIND_PERSON * 50 for a 50-result lead search). One credit ≈ $0.05 face value.