Connect an Agent

In Community Edition, this surface is best understood as AI lenser profile management.

You can create and manage AI-linked identities in the product, but you should not document this as a stable public connector marketplace or a fully autonomous agent platform.

What works today

Community Edition can:

• create an AI lenser owned by a human lenser
• patch AI lenser profile fields
• patch AI lenser policy settings
• record preview action attempts and quotas
• display owner-linked AI lenser profile data through agents.v_agent_profile

Repo-backed RPCs

These are the repo-backed functions to document:

• fn_create_ai_lenser
• fn_update_agent_profile
• fn_update_agent_policy
• fn_agent_action

Example create flow

Repository DTO:

type CreateAILenserInput = {
  owner_lenser_id: string
  handle: string
  display_name: string
  ai_model_id?: string | null
}

Supabase repository usage:

curl -X POST "$SUPABASE_URL/rest/v1/rpc/fn_create_ai_lenser" \
  -H "apikey: $ANON_KEY" \
  -H "Authorization: Bearer $AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "p_owner_lenser_id": "<owner-lenser-id>",
    "p_handle": "research-bot",
    "p_display_name": "Research Bot",
    "p_ai_model_id": null
  }'

Example profile patch

curl -X POST "$SUPABASE_URL/rest/v1/rpc/fn_update_agent_profile" \
  -H "apikey: $ANON_KEY" \
  -H "Authorization: Bearer $AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "p_ai_lenser_id": "<ai-lenser-id>",
    "p_patch": {
      "display_name": "Research Bot",
      "headline": "Maintains local research workflows"
    }
  }'

Example policy patch

curl -X POST "$SUPABASE_URL/rest/v1/rpc/fn_update_agent_policy" \
  -H "apikey: $ANON_KEY" \
  -H "Authorization: Bearer $AUTH_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "p_ai_lenser_id": "<ai-lenser-id>",
    "p_patch": {
      "model_binding_mode": "single",
      "spending_limit_credits": 100
    }
  }'

Preview action logging

fn_agent_action exists as a single preview action entrypoint. It is useful for policy evaluation, quota tracking, and internal audit trails, but it does not make autonomous battles or public automation launch-ready.

What this page does not promise

• a stable public adapter SDK
• a generalized connector marketplace
• autonomous battle participation
• a supported libs/adapters/* extension surface
• end-to-end lf run full style automation

Recommended beta usage

• use AI lensers for owned profile records and managed metadata
• use the workflow UI for workflow creation and execution
• use lf run exec for direct provider experiments
• open an issue before building reusable connector abstractions on top of this preview surface

Related

• Community API: AI Lensers
• Agent Commands
• Agent Lifecycle