Brand Logo

Agent-native prospecting · evidence first

Nooks turns a market question into a prospect list you can interrogate

Give your AI agent an ICP, a territory, and an evidence standard. Receive structured research with source context and review states without adding another seat-based workspace.

Inspect a sample result

Copying is step one; you still run and configure the skill in your agent.

INSTALLnpx -y @okki-global/okki-go-taroball
SAMPLE TASK
Find infrastructure software teams hiring RevOps, then show the evidence behind every match.
Nooks agent canvas with prospect evidence and review states

Workflow differences

Research stays visible from question to approved output

01ICP → observed signal → confidence

Keep observed facts apart from inferred fit

Firmographic and technographic criteria become explicit fields rather than hidden filters. Missing evidence stays missing; retrieval time, source context, and exclusions travel with the record so a reviewer can challenge the reason an account entered the shortlist.

02Draft → review gate → approved action

Put a person between research and outreach

Nooks can prepare context and a draft, while targeting, factual accuracy, suppression, opt-out handling, and sending remain separate approval decisions. That boundary supports sender reputation and regional compliance work without pretending that software alone makes outreach lawful.

Task evidence

Four prospecting jobs, each with an inspectable boundary

Find companies

Input
ICP, geography, exclusions, and required signals.
Output
A labeled account set with source context.

Limit: public evidence and provider coverage vary by market.

Verify contacts

Input
Approved accounts and target decision roles.
Output
Candidate contacts with provenance and verification state.

Limit: verification does not guarantee inbox placement.

Prepare context

Input
Account evidence, role mandate, and recent signals.
Output
A concise research brief with uncertainty retained.

Limit: inference requires human confirmation before use.

Create outreach

Input
Approved evidence, channel policy, and tone.
Output
A review-ready draft, not an automatic send.

Limit: recipients, suppression, and legal basis remain operator decisions.

Auditable preview

A sample result carries its own review trail

SAMPLE

Prospect result

Sample Account A

ICP match: infrastructure software. Signals: RevOps role posted; CRM migration mentioned.

Sample Account B

ICP match: finance operations. Signals: territory expansion; sales operations hiring.

Four-state runbook

Copy is not install; install is not a first result

  1. 01

    Copy

    npx -y @okki-global/okki-go-taroball

    Prerequisite: access to the intended agent terminal. Success: the exact command is in the clipboard. If copying fails, select the code manually. Next: run it.

  2. 02

    Run

    Read the package prompt before allowing execution. Success: the installer completes without a package or permission error. If the runtime blocks npx, stop and follow its package policy. Next: configure.

  3. 03

    Configure

    Add only credentials required by the chosen workflow and keep keys in the runtime secret store. Success: the agent can enumerate the skill without exposing secrets. If authorization fails, reduce and reissue the scope. Next: test.

  4. 04

    First result

    Begin with a narrow ICP and request evidence fields. Success: a structured sample appears with provenance and review states. If fields are missing, adjust the task or provider configuration before expanding volume.

Security ledger

Trust labels describe what still needs verification

Data sources

  • Config-dependent Provider choice and geographic coverage.
  • Verify locally Source freshness and field provenance in each result.

API keys

  • Operator-controlled Secret-store location and permission scope.
  • Not disclosed here Third-party provider retention.

Runtime behavior

  • Verify before run Package prompt and requested access.
  • Config-dependent Connectors, exports, and external actions.

Retention and deletion

  • Operator-controlled Local result handling.
  • Verify downstream CRM and provider deletion behavior.