Point your AI agent at the network

You are setting up a Reqport responder integration for my organization. Reqport is a network for receiving and answering authority data requests in the ARM family (for example a business-relationship check, a free-text note and attachments, and a crypto transaction-history request); request content is decrypted only inside a confidential TEE for our org. The family, the authorityArmEnabled gate, and where auto-release vs human approval is decided are at /developer/exchange/arm. Do this end to end and report at each step:

1. Install the CLI (binary is `qp`):  npx @reqport/cli@latest --help    (or: npm i -g @reqport/cli)
2. Authenticate:  qp login    This opens a browser where an org admin approves and a per-environment API key is minted for the CLI (no secret is pasted by hand). Use --env sandbox for testing and --env prod for live.
3. Verify:  qp doctor   and   qp whoami
4. Discover requests addressed to us:  qp --env sandbox requests list
5. Read one (decrypted in the TEE for our org):  qp --env sandbox requests show <REQUEST_ID>
6. Answer a business-relationship check:  qp --env sandbox respond <REQUEST_ID> --has-relationship true   (we are their customer)   or   --has-relationship false   (no relationship). For a crypto transaction-history request, answer with a camt.053-CA statement JSON:  qp --env sandbox respond <REQUEST_ID> --statement statement.json
7. To drive this from your own tools, run  qp mcp  to start a stdio MCP server exposing the same operations as tools, with the exchange contracts available as context.

Start in --env sandbox (safe, seeded with synthetic requests). Confirm the full discover → read → respond loop works there, then switch to --env prod to go live. If a step needs an org admin (approving the CLI pairing), tell me exactly what to click.