Status#

Guards waiting on a chain, not on an SDK#

Most of these are built here and refused here, with a named diagnostic — they never reach the user as a bare node error, and nothing is spent on the way to failing. --nft is the exception: it is built, signed and broadcast like any other launch, and its diagnostic fires on the chain's answer rather than instead of it. A -25 back names the candidate causes without picking one, and says to check before resending — because unlike a refusal made here, it does not settle whether the fee went. (An NFT is charged the parent's idimportfees, 0.02, not the 200 a currency registration costs.)

They were refused because the SDK could not express them. That stopped being true on 7 August, when every one of these closed upstream; the pin caught up on the 22nd. What stands between them and working is no longer a dependency, it is evidence: each is a money path, and the only tests that exercise one spend real VRSCTEST against a public node. So the guards stay until somebody removes one and watches it land.

	closed upstream by	what pecu does today	what removing the guard takes
Mint / preconvert / convert to a VerusID — `build_conversion` wrote every recipient as `PubKeyHash` and dropped the kind, so an i-address would have paid the R-address sharing its hash	#115 — `convert.rs` now maps `AddressKind::Identity` to `Destination::Identity`	refuses, and says paying a primary address is not the same thing	pay an identity a token and read it back off the chain as the identity's, not a key holder's
Token send to a VerusID — same root cause, so an identity could hold native coins but not tokens	#115	refuses with the native-vs-token asymmetry spelled out	as above; the two share a code path and should be proven together
NFT launch — the identity output omitted the tokenized-control recovery destination, so consensus derived a different script	#111 — `identity_primary_script` now takes `has_tokenized_control`	`--nft` builds and broadcasts it; a `-25` back names the candidate causes rather than picking one	launch one, and check consensus accepts the identity output rather than `bad-txns-failed-precheck`
NFT launch fee — charged `currencyregistrationfee` (200) instead of `idimportfees` (0.02)	#112	pins the right fee itself	confirm the fee the flow now reads matches the one pinned here, then stop pinning it
NFT definition shape — `NFT_TOKEN` on a `token()` could not express a valid one	#113 — `CurrencyDefinition::nft()` exists	sets all five fields by hand	decode a launch built by the constructor against one built by hand; they should agree field for field
Seeded contributions at launch — the builder emitted no output funding a declared contribution, and the notarization in the same transaction said the reserves were zero	#129 — the daemon's eighth output is built now	refuses `--contribute` and points at `preconvert`	launch a basket with a seeded reserve and confirm the reserve actually holds it at the start block

Two rows left this table rather than moving down it.

The expired commitment (#114) is done. CommitmentStatus::Expired exists and resume matches it by name, before any broadcast — where the old string match on expiring-soon could only fire after one had already been attempted, and the two states need opposite actions. The string match is kept as a fallback for the same rejection arriving unclassified.

Two payments in one block (#118) needs no guard removed, because pecu never had one — the limitation was simply true. funding now reads the mempool and withholds coins an unconfirmed transaction already spends, so a second payment should build against different coins. Unproven here: it wants two sends in one block, watched.

Built but unproven#

Paths that exist and have never been exercised against the chain. They are not claimed as working:

pecu key --history on a key with more than one revision. The single revision case is covered; nothing has rotated yet.
Multi-signature identities. send, mint and the whole lifecycle sign with one key and refuse an m-of-n identity by name. The air-gap trio is the natural home for this, once it learns identity inputs.

Refused because nothing has built it#

Different in kind from the upstream-gap table further up: nothing closed upstream, and no chain evidence is pending. The code was simply never written, and deliberately is not being written now.

Tokens and identity-held funds through the air gap. pecu send has three modes and pecu plan send carries one of them. It accepts --currency and --from-identity and refuses each by name — pecu::plan_has_no_token_path and pecu::plan_has_no_identity_path — before it opens a socket. prepare_unsigned_send is the only SDK builder that returns a partial transaction; every token and identity builder signs as it builds, so there is no unsigned form of either payment to carry to the offline machine. The partial format is not the blocker — a partial is started with whatever outputs it is handed, and it has an input kind for a pay-to-identity input already. What is missing is a builder, not a byte shape. pecu send moves both, signing on the machine that talks to the node.

Parked#

feat/m8-login-vdxf — id login and VDXF publish/read, deliberately kept off main. Well behind now; needs a rebase before it is worth looking at.

Not in scope (yet)#

Shielded/z-address operations, the currency launch wizard, marketplace offers, and a live watch dashboard. They come after the transparent and identity paths are solid.