Agent-to-agent interface (A2A)

Another AI agent can use the A2A protocol to consult Laurent's public profile, skills and suitability for a mission.

You are talking to an automated conversational agent (AI), not directly with Laurent Lepczynski.

This interface answers incoming requests in read-only mode. It does not negotiate or make commitments on Laurent's behalf.

Interface status

Active

Public consultation is open, with rate, concurrency and expiry limits. No autonomous actions.

Available capabilities

Public profile lookup

public_profile_lookup

Returns the versioned public professional profile (experience, expertise, languages), without contact details or internal notes.

Sourced public profile question

public_profile_question

Answers a specific question from the published public knowledge base, with cited sources and a confidence level; declines rather than inventing what isn't covered.

Limits and restrictions

Bounded public access

No key is required for the two public capabilities. Introduction requests remain closed.

  1. Fetch the agent card : GET /.well-known/agent-card.json.
  2. Send a text message using POST https://www.laurentlep.com/a2a · message/send. The messageId field allows a request to be retried without running it again.
  3. Read the immediate result, or retrieve it with tasks/get from the same network address before it expires.

Example: suitability for a mission

{
  "jsonrpc": "2.0",
  "id": "mission-1",
  "method": "message/send",
  "params": {
    "message": {
      "kind": "message",
      "role": "user",
      "messageId": "mission-1-unique",
      "parts": [
        {
          "kind": "text",
          "text": "A mission involves leading an AI transformation and a sales team. Which public experiences from Laurent's background match, and what still needs to be confirmed?"
        }
      ]
    }
  }
}

Answers rely on public sources. Suitability is an interpretation of the professional background, without a guaranteed score or commitment. Information missing from the sources must still be confirmed. Questions are limited to 900 characters.

Single-turn text exchanges and structured responses. Files, push notifications and multi-turn conversations are not supported. A completed task cannot be canceled with tasks/cancel. Legacy skillId/params calls remain available.