Skip to main content

What is the difference between resolve and enrich?

Resolve answers “who is this”: it turns an email, a LinkedIn URL, a name with a company, or a domain into one stable Agentic ID, and says whether the ID was known, is new, or needs review. It returns the ID, not facts. Enrich answers “what do you know about them”: it fills in fields such as title, company and work email. Each value comes with a level and a source.

Am I charged for an empty result?

No. You are charged for results delivered, and cost.charged_credits in every answer says how much. Specifically:
  • Validate: an unknown or rejected verdict is not charged.
  • Enrich: nothing found, nothing charged.
  • Find, discover, lookalike: charged per person or account delivered.
  • Discover with count_only: true is not charged.
  • A request refused with a 4xx error is not charged.
See Credits.

What do quoted_credits and charged_credits mean?

cost.quoted_credits is the most the call could have charged. cost.charged_credits is what it did charge. Charged is never more than quoted, and never more than your max_credits.

Why does a result say stopped_early?

The search stopped at one of Agentic ID’s own limits on paid sources, not at your cap. The answer is a normal 200 and you are charged only for what was delivered. See Errors.

Why did I get fewer companies than I asked for?

Discover sets fewer_than_asked with the reason. One call reads a limited number of result pages and drops candidates that fail a check. excluded lists why candidates were left out, with counts. Ask again for more: what was already delivered is skipped. Narrower filters fill a call better.

Why are contact values masked?

The API returns contact values (emails, phones, LinkedIn profile URLs) masked today, for example a******@acme.example. Names, titles and company data are not masked. See Data and privacy.

What do verified, corroborated and claimed mean?

They are the level of a value: how sure Agentic ID is about it. See Verification.

Is confidence on a resolve the same as level?

No. confidence says how sure Agentic ID is that your identifier points at that ID. It does not say how verified the ID’s facts are. For that, call GET /v1/facts/{id}.

Why does resolve say known when I have never looked this up?

outcome: "known" means Agentic ID already had this ID. It does not mean your workspace held it. The summary says which: “already had this ID; it is new to your workspace” or “your workspace already holds this ID”.

Why do facts or ledger return 404 for an ID I have?

Facts and ledger are shown only for IDs your workspace holds. An ID copied from elsewhere returns 404 id.not_held. Resolve the person or company first.

Can I use it without Claude Code?

Yes. Four ways:
  • The REST API from any language. See the API reference.
  • The MCP server with any MCP client that can start a local server.
  • The CLI from a terminal or a script.
  • The SDK, once it is published.

Do I need an API key?

No. You can sign in with your account instead (OAuth 2.1 with PKCE). Use an API key for scripts, CI and headless runs. When both are set in the MCP server or the CLI, the key is used. See Authentication.

Is it safe to retry a request that timed out?

Yes, if you sent an Idempotency-Key. Send the same request with the same key: it is run and charged once. Without a key, check GET /v1/credits for a charge before you send it again. See Idempotency.

Can I look up personal emails or phone numbers?

No. Agentic ID works on work identities and companies. Validate returns rejected for personal, disposable and malformed addresses, and does not charge for them. The MCP server and the CLI hold back personal emails and phone numbers before anything is sent.

How do I see what I spent?

GET /v1/credits returns the balance, spend today and in the last 30 days by tool, the latest grants and charges, and the IDs delivered with what each cost. Times and “today” are in UTC. In the MCP server use agentic_id_workspace. In the CLI use agentic-id workspace.