[DOCS] client-sdk/guides: migrate the SDK guides to 0.7

[alexandrecarvalheira]

Stacked on #53. First batch of the code sweep: the seven guides that teach the SDK API, where wrong code costs the most.

Verified against the published SDK docs, not just my rename table

You were right to push back on the scripted rename. I pulled https://cofhesdk.fhenix.io/llms-full.txt and checked each page against it, and the script had produced five real bugs that source-checking alone would not have caught:

.withPermit(permit) became .withACP(ACP) in four places, because the regex capitalised the variable, not just the type. The correct call is .withACP(acp). A reader copying that example gets an undefined identifier.

It also rewrote the literal error strings 'Permit is expired' / 'Permit is not signed'. That one happened to be right, but only by luck: I confirmed against acps/validation.ts that 'ACP is expired' and 'ACP is not signed' are the actual strings in 0.7.1.

What the official docs confirmed: the three withACP overloads, the exactly-one-of rule on decryptForTx, the { ctHash, decryptedValue, signature } return shape, the [...hashes, signature] batch, the Encryptable to Solidity table row for row, the 2048-bit limit, and // defaultACPExpiration: 60 * 60 * 24 * 30 verbatim.

What changed

encrypting-inputs is rewritten. Every example was wrong: execute() now returns [...hashes, signature], which is inputs.length + 1 elements, and setConsumingContract is required before execute() exists at all. Also documents the trap with no compile-time signal, that the consuming contract is whichever contract runs FHE.asEuint* and not necessarily the one you call, and adds setSecurityZone, which was never documented.

writing-encrypted-data moves off the deleted InEuint32 struct to externalEuint32 plus a bytes proof. The hand-written viem ABI changes from the (ctHash, securityZone, utype, signature) tuple to bytes32 and bytes, and assertCorrectEncryptedItemInput is dropped (zero files in 0.7.1).

error-handling had two error codes that no longer exist. PermitNotFound and PermitInvalid are ACPNotFound and ACPInvalid, and the enum gained ACPExpired, ACPRevoked, ACPDenied, ACPRequired, so callers no longer infer the condition from the message. Added ConsumingContractUninitialized.

decrypt-to-view, decrypt-to-tx, client-setup take the renames.

Housekeeping

Drops the {#validating-permits} anchor I added to the ACP guide in #53. It existed only to keep this page's deep link alive across the rename; that link now points at the natural slug, so the workaround is gone rather than left to rot.

Vale and lint-docs.py exit 0 on all eight guides. mint validate and mint broken-links --check-anchors --check-redirects pass.

🤖 Generated with Claude Code