What the October 2026 research rounds found and decided, what has been built, merged and released since, the roadmap, the corrections the research made to itself, and the decisions still open. Every status on this page was checked against the code (zkas-rusty main d6e22ec5, zkas-wallet 1.0.52-rc28) and the published releases on 3 November 2026 (first written 28 October). Companion to the Idea & Capability Map v2 · Core Map.
Status keys: live running for users · in main merged, not yet released or deployed · branch built and tested, not merged · spec designed only · not started.
Live since 26 Jul 2026 (genesis b63f7fe8…), merged-mined with Kaspa on kHeavyHash. Two hard forks since launch: NU1 (DAA 757,000, 4 Aug) and the security fork (DAA 6,983,500, ~15 Oct), which also ended the 5% development share.
Post-fork rules, the four remote-crash fixes, the N1 hardening, the scan engine and uniform sends (both off by default). Published for self-hosters on zkas.info/downloads/node; the public nodes still run an earlier build.
Windows, macOS, Linux, Android and web. Syncs from signed public chain files on a CDN, and on wallet.zkas.info can run the wallet engine inside the browser. Downloads · signed Radicle releases.
GitHub suspended the account on 23 Oct and GitLab blocked it on 24 Oct. Since 26 Oct all 11 repositories live on Radicle, browsable at zkas.info/code; no account can take them down.
| Component | State | Detail |
|---|---|---|
| Consensus: post-fork rules (anchor naming, bound merge-mining ZKM1, state root, no dev share) | live | Security fork at DAA 6,983,500 closed fake anchors (ZK-01), double credit for one mining solution, block-level malleability, peer-supplied accrual and the coinbase stall. |
| Node release N1: no nullifier-import cap, fee checked before the proof, batched pre-verify, verify budgets, ban for invalid proofs, era-tagged nullifiers, deduplicated pruning proof | in main | Merged 27 Oct (fe049dee) and released in node v1.0.13. Includes the fix for the batched verifier that rejected every post-fork bundle. Not yet on the public nodes. |
| Signed public chain files + chain gateway | live | Since 31 Oct–1 Nov: the shielded chain as manifest-signed 12-hour segments on a CDN (zkas.info fallback) and a public gateway at zkas.info/chain-gateway. A phone synced a 62-day wallet in 217 s from the CDN alone; every segment is checked against the signed manifest and the node's frontier. |
| Run in this browser | live | Since 1 Nov on wallet.zkas.info: the wallet engine, compiled to WebAssembly, scans inside the browser, so no server sees the viewing key. Receive, balance and history today; sending from the browser is next. |
| Ready-note ladder (no wait on your own change) | in main | walletd efd73fcc: a payment that would leave fewer than 3 settled coins returns its change as several coins. |
| Multi-party atomic bundles | blocked | In code behind a default-off flag, but walletd refuses them since the security fork because the builder does not set the mandatory anchor_block (about 0.5 dev-month to fix). Every card that rests on them is marked blocked. |
| Chunked pruning-point proof transfer | in main | 27 Oct (25a9c5aa…d2c97763), negotiated by a services bit with fallback for older peers. |
| Multi-wallet scan engine (phases 1 + 2) | in main | Merged 27 Oct, off by default. In a shadow run beside the hosted daemon it used 0.007 cores and 4 GB of chain data instead of 0.74 cores and 671 GB, migrated 27/27 real wallets without a rescan and showed no unexplained difference. Not serving users yet. |
| Uniform transactions, Stage W (every send the same 4-action shape) | in main | Merged 27 Oct, opt-in. Raises the crowd a timed payment hides in from a median of 1 to ~155 (±5 min) and ~926 (±30 min). |
SF-L (shielded lock_time enforced) and NU-A (burned fee floor, invalid proof makes a block invalid, action cap, tree-era rollover, state root v2) | branch | core/sf-l-nu-a, devnet-proven, every activation set to never. Consensus changes: need sign-off and an activation height. |
| p2p v11: encrypted transport (ZKT2) + Dandelion++ | branch | core/p2p-v11. On a devnet it cut a spy node's chance of naming the sender from 100% to 18–52% for +0.12 s latency. The Tor submission relay that removes the rest is not built. |
| Tagged channels (detection without trial decryption) | branch | core/tagged-channels, devnet only: 0 trial decryptions, 0.014 CPU-s per chain-day vs 2.84 core-s today. |
| Hosted wallet daemon | live | Memory budget, shared leaf stream and nullifier interner: per-wallet memory 1,653 → 816 MiB, live cost 2.86 cores / 72.7 GB → 0.74 cores / 22.9 GB. |
| Explorer, solo merged-mining pool, self-host bridge v1.0.12, OTC desk, payment gateway, SDK 0.2.0, docs | live | Bridge v1.0.12 (24 Oct) is the first self-host bridge that pays after the fork. |
| Private coinbase, Kaspa-anchored finality, transaction format v2, light-client protocol, post-quantum notes | spec | Specified on 21 Oct with numbers; see roadmap. |
| Public supply page, third-party light client, iOS app, notarised macOS build, public testnet | not started | See open decisions. |
| Date (UTC) | Event |
|---|---|
| 29 Jun 2026 | Project starts as a rusty-kaspa v2.0.1 fork; renamed FireCash (3 Jul) and ZKas (14 Jul). |
| July | About six test chains and relaunches: 10 → 1 blocks/s (10 Jul), rollback after a chain freeze (20 Jul), bridge reset (25 Jul). |
| 26 Jul | Current mainnet genesis b63f7fe8…. |
| 4 Aug | Hard fork NU1 at DAA 757,000. |
| 2–6 Oct | Core security audits; no-fork fixes released for the remote node kill, forged import root, history-backfill wedge and reorg-crash nullifier loss. |
| ~15 Oct | Security fork at DAA 6,983,500; the 5% development share ends (≈16M ZKAS, ≈4.8% of supply, went to the published dev address before it). |
| 18–21 Oct | Hosted wallet send outage (~2.5 days, node/walletd version skew on the fork's anchor_block) and out-of-memory restarts; both fixed and deployed by 22 Oct. |
| 19–20 Oct | Adoption research (two rounds, 22 passes) and the 44 product write-ups; consensus roadmap (16 lenses). |
| 21 Oct | Deep round (measurements, errata), 14 core specs with reconciliation and fixes, forks study. |
| 22–25 Oct | Implementations: N1, SF-L/NU-A, Stage W, p2p v11, tagged channels, scan engine phases 1–2; experiments E1–E11. |
| 23–26 Oct | GitHub suspended, GitLab blocked; code moved to Radicle with zkas.info/code links. |
| 27 Oct | N1, scan engine and Stage W merged; node v1.0.12; wallet rc17 (all platforms) and Radicle releases. |
| 28 Oct | Wallet rc18 with the post-fork node; all public links moved to zkas.info. |
| 29 Oct–1 Nov | Wallet sync measured end to end; frontier almanac, coinbase fast path, signed chain files on a CDN, the chain gateway and browser mode; wallet rc19–rc28; node v1.0.13 published. |
| 3 Nov | Every idea-map card re-checked against the code: 18 removed, 3 added, statuses corrected (see §14). Code links now open on the public Radicle seed iris. |
Each round was run as parallel research passes, then reconciled; later rounds checked and corrected earlier ones.
| Round | Date | Scope | Main output |
|---|---|---|---|
| Security audits | 2, 5–6 Oct | Consensus, supply, anchors, reorg/pruning, mempool, AuxPoW headers, IBD, wallet/signer/walletd, bridge | Findings fixed by releases and by the security fork (§04). |
| Adoption, rounds 1–2 | 19–20 Oct | 13 lenses (privacy season, traders, lawful verticals, capital, miners, exchanges, geography, competitors, contrarian, machine money, architectures, inflows, integrations) + 9 round-2 passes (red team, supply, content, listing kit, gift loop, x402, grants, Kaspa relations, verification) | An ordered plan: disclose and fix first, depth before more listings, gift loop as the first product (§10). |
| Product write-ups | 19 Oct | 44 files, ~300 named lawful products | All run on shipped primitives; the missing pieces ranked (§11). |
| Consensus roadmap | 20 Oct | 9 core lenses + 7 privacy lenses + red team | Seven hazards to fix now, a no-fork program, then NU-A and NU-B (§07, §08). |
| Deep round | 21 Oct | Measurements, size, scanning, anonymity set, crypto core, security, threat model, interop, errata | Real numbers for the chain, and a list of what the earlier round got wrong (§05, §12). |
| Core specs | 21 Oct | 14 specs + reconciliation + fixes | Eight release trains, ≈73.5 core dev-months (§06). |
| Forks study | 21 Oct | Oblivious sync, succinct chain, throughput, native DeFi, next pool, Kaspa-native, programmable notes, moonshots, the stack | A stack and order to ≈2030 (§09). |
| Wallet daemon and client | Jul–Oct | Scaling, incidents, engine, diagnostics | 39 incidents documented, all note-loss classes fixed (§05b). |
core/sf-l-nu-a for NU-A; N1's limits reduce the cost meanwhile.lock_time is signed but not enforced for shielded transactions (a trap for any contract that relies on it). SF-L fixes it; a census of 551,789 mined shielded transactions found none with a non-zero lock_time, so nothing is stranded.Deep round, 18–21 Oct, unless stated.
| Quantity | Value |
|---|---|
| Load | 3,435 transactions and 68,659 actions per day; 1.05% of block capacity used; 48.9% of new notes are coinbase |
| Size | 3,156 B per action + 2,869 B fixed per bundle; smallest payment 9,275 B; a block holds 76–78 actions (250 KB) |
| Proving | ≈ 0.27 + 0.531·n seconds for n actions |
| Verifying | ≈ 16 + 0.63·n ms per bundle; the whole network's verification load is 0.11% of one core |
| Wallet scan | Trial decryption 43.4 µs per action; 82% of scan CPU was one duplicated coinbase commitment (fixed, shipped in rc17) |
| Storage | 14.98 M note leaves; ≈ 0.96 GB/day archival growth |
| Hashrate | ≈ 76–78 PH/s merge-mines ZKas (≈ +46% revenue over Kaspa-only mining for those miners) |
| Anonymity | A payment whose time is known to ±5 min matches a median of one other transaction (45% unique); one uniform shape lifts that to ~155 |
| Live-chain attribution (public data only) | 85% of transactions attributable to three automated senders; 92.4% of new value-bearing notes have a public owner (mostly coinbase); every transaction pays the exact default fee |
| Change | Before | After | State |
|---|---|---|---|
| Hosted daemon memory budget + shared stream + nullifier interner | ~120 wallets, 58 GB + swap, OOM every few hours | flat 15–23 GB; 816 MiB per wallet | live |
| Send with a large wallet (subtree cache) | 29.4 s | 0.36 s | live |
| Restore (birthday fast-sync) | 45 min | 20 s | live |
| Second device with the same viewing key | full scan | 0.032 s | live |
| Decode per chain-day (commitment once) | 7.03 s | 3.72 s | live |
| Phone restore, 30 days of chain | 24.7 min | 7.5 min | live (rc17) |
| x86-64 assembly backend | portable | 18–34% faster | live |
| Shared-leaf checkpoints (v11) | 84 MB per wallet file | 34–37 MB | built, off (one-way format change) |
| Multi-wallet scan engine | 0.74 cores, 671 GB | 0.007 cores, 4 GB (shadow) | in main, off |
39 incidents from July to October are documented with cause and fix. Every class that could lose a balance (pruned rescans, notes deleted at submit, stranded notes, checkpoints wrongly rejected) has a deployed fix and a recovery path. The client gained on-device diagnostics with one redaction layer shared by the JavaScript, Rust and Kotlin code (no private data found in test logs on Linux, Windows, macOS or Android), and 115 client audit findings were fixed.
| Bug found by the specs round | Effect | State |
|---|---|---|
| Batched verifier rejected every post-fork bundle | No batching margin against verification DoS | fixed in main |
| Invalid-proof transaction in a block silently dropped | Nodes verify garbage for free | rule on branch + N1 limits in main |
| Coinbase note commitment computed twice in every wallet | 71% of a chain-day sync | shipped in rc17 |
| Train | Content | Kind | State |
|---|---|---|---|
| 0 · N1 | Nullifier import without a cap, DoS ordering, shared flag mask, stream v2 RPCs, era byte, standardness | Node, no fork | in main |
| 0 · W1 | Scanner and sync fixes, Stage W, hot purse, engine v2 | Wallet | sync fixes live Stage W + engine in main |
| 1 · p2p | ZKT2 + Dandelion++ with an onion relay, compact blocks, conflict alerts, Kaspa-finality shadow mode | Node, no fork | transport + D++ on branch |
| 2 · SF-L | lock_time enforced for shielded transactions | Soft fork | branch |
| 3 · NU-A (Q1 2027) | Burned fee floor, invalid proof block-fatal, action cap, dormant tree-era rollover, fixed coinbase extra data, state root v2 | Hard fork | branch |
| 4 · FC-K (Q2–Q3 2027) | Kaspa-anchored finality: a Kaspa view, a finality floor, conflict tracker | Fork-choice rule | spec |
| 5 · NU-B (mid-2027) | Transaction format v2, sighash v2, anchor epochs, exact fees, recoverable V3 notes, private mining rewards (Design K), state root v3 | Hard fork + key switch | spec |
| 6 · NU-C (2028+) | Detection tags, anchor-depth cut, capacity raise, each on a measured trigger | Hard fork | spec |
Binding decisions from the reconciliation: anchor epoch 120 blocks with depth 600 (no depth cut before NU-C); one flag layout; the fee floor is burned and the tip goes to the miner; private rewards can lapse but never expire; Kaspa-anchored finality is a fork-choice rule, not a consensus rule.
| When | What |
|---|---|
| Now (no fork) | Node v1.0.13 with N1; wallet releases that turn on Stage W; message signing without publishing the viewing key; explorer stops emitting peer identities at the source; supply page. |
| Months 1–5 | Shard roots and a sparse stream for faster phone scans (≈4–5×); a reorg-safe wallet tree; the wallet checks the state root against proof of work; Kaspa-witness index and alarm; out-of-band delivery. |
| Months 3–6 | p2p v11 (encrypted transport + Dandelion++); mining nodes stop including their own transactions; light mode by default. |
| Q1 2027 | NU-A, with SF-L at the same or an earlier height. |
| Months 6–9 | Private information retrieval for catch-up. |
| Mid/late 2027 | NU-B: format v2, private coinbase, ML-KEM hybrid encryption for post-quantum safety of received notes. |
| 2028+ | NU-C on triggers; a STARK-based post-quantum pool, recursive light-client checkpoints and a nullifier accumulator on the 5-year track. |
Rejected after analysis: 10 blocks/s, view tags, fuzzy message detection, proof aggregation on the validity path, revocable viewing, association sets, trusted hardware, a full multi-asset (ZSA) pool now, and any new circuit design of our own.
Constraints the research changed: the coinbase recipient does not have to stay public; fees no longer all go to the miner (the floor is burned); a second asset pool fits every rule; a public testnet is required; a RandomX parent chain is deferred.
33 leaks were tabulated with severity and fix on 20 Oct. The critical ones:
| Leak | Fix | State |
|---|---|---|
| Coinbase recipients are public (most new value has a known owner) | Private mining rewards, Design K (NU-B); one-time coinbase addresses until then | spec |
| A merged block links to its Kaspa payout address | Payout hygiene now; private coinbase later | spec |
| Explorer published peer and payout joins | Filtered at the web server since 20 Oct; to be removed in the explorer source | stopgap live |
| Hosted wallets hold viewing keys and history | Light mode by default, server-side key deletion, the scan engine | engine in main |
| Automated senders have fingerprints (shape, fee, timing) | Uniform transactions (Stage W), Poisson merges and payouts | in main, opt-in |
| Network origin of a transaction | ZKT2 + Dandelion++, then a Tor submission relay | branch |
| Message signing publishes the full viewing key | An address proof without the viewing key | not started |
The forks study (21 Oct) asked what a mandatory-privacy, Kaspa-merged BlockDAG could do that no other chain does, across eight topics: oblivious sync that ends scanning, a succinct chain, payments that feel instant, private DeFi in consensus, the next shielded pool, Kaspa-native designs, notes that carry hidden programs, and contrarian moonshots. It ends in one recommended stack. That stack is an alternative to the baseline trains in §06, not yet adopted: two experiments decide between them.
Cost ≈194 core dev-months (vs ≈73.5 for the baseline plus 34–44 for pool Z), ≈3–4 years at 4–5 engineers. The two experiments that decide the most: E1, posting ZKas data on Kaspa mainnet plus agreement with Kaspa's core developers (Based ZKas, or keep FC-K), and E2, zero-knowledge hiding in the STARK prover (pool Z, or a Halo 2 fallback). Neither has been run.
Built on core/tagged-channels (devnet only). A two-node devnet run found payment and change with zero trial decryptions and restored both wallets from seed; 0.014 CPU-s per chain-day against 2.84 core-s for trial decryption; outputs 18% smaller. Open: private lookup is still too slow (3.29 s and 1.14 MB per query at 4 GB), restore without a birthday works only while the chain is young, and fetching a record after a tag hit reveals which outputs are yours, one of three fixes is needed before any mainnet plan.
| Item | State, 28 Oct |
|---|---|
| Explorer peer-identity leak | filtered |
| OTC raffles feeding the listing fund; listing campaign page | live |
| Self-host bridge that pays after the fork (v1.0.12) | live |
| Homepage wording and development-share disclosure | not changed, owner decision |
| Supply page, objections / Orchard / security / pools pages | not built |
| Gift link, x402 facilitator, BTCPay plugin, Cashu mint, light client | not built |
| Outreach (market makers, listings, grants, reviewers) | No record of anything sent |
The 44 product write-ups (about 300 named lawful products: payroll, gifts, merchants, miner distribution, disclosure, remittance, receipts, bounties, estates, mutual aid and more) all run on primitives that already ship: fresh diversified addresses, the 512-byte memo, payment proofs, viewing keys and the spend wait. None needs a new circuit or a fork.
The deep round re-measured what the 20 Oct round estimated. Any number below supersedes the earlier one wherever it appears.
| Earlier claim | Corrected |
|---|---|
| ~1,230 spends per day | ~69K actions per day (the explorer counter restarts with the service) |
| ~97% of new notes are coinbase | ~49% |
| A block holds ~39 actions / 125 KB | 76–78 actions / 250 KB |
| Nullifier cap ~4 years away | 1.2–2.4 years (now removed by N1) |
| Note tree full in 3.6 years | ~1.8 years at full blocks |
| Shard roots make phone scans 15–20× faster | ≈4–5×; download dominates |
| walletd checks the state root | It does not; it trusts the node |
| The fee is checked before the proof | It was checked after; N1 moved it before |
| "Effective anonymity set 8–15"; a 16-axis privacy scorecard | The first measures issuance concentration, not anonymity; the scorecard was an invented metric and must not be used in any claim |
| V3 notes are "a proving-key change" from an audited circuit | A different circuit with a new verifying key; no audit found, circuit-fork lead times apply |
Every card on the idea map was read and checked against today's code and the research. 95 cards changed (statuses, numbers, mechanisms), 3 were added for things that now ship (browser mode, encrypted wallet chat, private classifieds), and these 21 were removed (18 in the first pass, 3 merged into other cards in the second):
| Removed card | Why |
|---|---|
| Show up: a no-show penalty nobody ever holds | The card's own dependency defeats it: nothing is locked, so a no-show can spend the pledged note before dinner and the venue collects nothing. It also rests on multi-party bundles, which walletd refuses since the security fork (code check, Nov 2026). A penalty that cannot be collected is not a product. |
| Memory-budgeted hosted wallet: cgroup-aware walletd | An internal operations fix for the hosted daemon, not a capability anyone builds on or chooses ZKas for; it is recorded on the progress page. Its headline also overstates the result: the 30 Oct review measured eviction thrash (p90 wallet open 24 s), which the multi-wallet engine card addresses. |
| Post-loss mutual: a syndicate with no pooled capital | Members commit to nothing in advance (the card says a refusing member 'suffers nothing on-chain'), so it is a pass-the-hat with atomic delivery, not cover. Mutual aid without a pot is already the push-only group money card. It also rests on multi-party bundles, refused by walletd since the security fork. |
| Expiry repo: a money market where default is the absence of an action | Not a working loan. The unwind cannot be pre-signed (anchor rule), so a lender keeps collateral that rose and a borrower abandons collateral that fell: each side holds a free option against the other. On a single-asset chain ZKAS-against-ZKAS repo is meaningless, so the collateral's value is off-chain anyway. Rests on multi-party bundles, refused since the security fork. |
| KAS-leg conditionality + 2-of-2 ZKAS notes: escrow, oracles & DLCs across both chains | Dominated: by its own text it is incentive-atomic only and the KAS bonds publish stake sizes; DLCs after tx format V2 and the forks study's policy notes do the same job privately, and the forks study already dropped this track. |
| Kaspa-rollup DeFi (AMM / lending) with wZKAS: ZKAS as the private settlement asset | Privacy exists only at entry and exit, so this is Kaspa-rollup DeFi holding a wrapped token rather than a ZKas capability; the part that is ours (a wrapped ZKAS) is the wZKAS card. |
| Trustless KAS↔ZKAS peg: burn-in via SPV reserve, peg-out blocked | The card's own verdict is 'keep the code, don't build on it': BRIDGE_ENABLED is false in consensus (checked in code), the deep round recommends deleting the dormant peg-in, and the live design is the Based ZKas peg in the Based ZKas card. History belongs on the progress page. |
| Based-rollup shielded-KAS + ZKAS-fee layer | Superseded by the Based ZKas card (ZKas itself based on Kaspa, shielded KAS as an asset in ZKas's one tree), and it competes with ZKAS-as-asset by the card's own account. Named in the Based ZKas card as the alternative considered. |
| Hashrate-bonded operator / relayer set | By its own text hashrate identity is rentable and only the KAS bond does any work, so it is a plain bonded set worth about $30 of slashable stake today, and the forks study shelves its other form. It is an implementation detail of the Aux-PoW bridge's Class P path, not an idea of its own. |
| Recursive validity proof: O(1) light node | Its own verdict: superseded by pool Z's chained epoch proof and a Kaspa-side root beacon; recursion never goes on the 1-second validity path. Pool Z has its own card. |
| Withholding tribunal: slashing a pool that hides blocks | Slashing cannot be made fair: by the 21 Oct correction 53% of tags never resolve by design, honest orphaning is indistinguishable from withholding, and the card concedes a patient attacker stays under any threshold. The useful part (a non-slashing withholding alarm) is part of Kaspa-anchored finality's monitor. |
| Dark book-build: sealed bids in private money, caps enforced in public script | The seller custodies every bid and the card itself calls the bond 'theatre' unless it exceeds the raise; it also needs an unbuilt clearing circuit and covenant. Its one addition (pre-registered notes make double-bidding impossible) moves into the sealed-bid auction card. |
| Sealed blob lane: binding data that nobody can read yet | 'Binding while hidden' is the revealable notarization card; what this adds is a slashing covenant for a 512-byte-per-spend trickle, which the card itself says can carry keys and hashes, never data. |
| PoW-anchored randomness beacon / fair lottery | A shielded state root is not unbiasable randomness (a miner can withhold to grind it) and the VDF that would fix that is unbuilt; block-hash lotteries exist on every chain, so this is not a ZKas capability. It was rated 1/1 on privacy and growth. |
| Arity subsidy: paying for the public shape of a private transaction | Works against the privacy plan now in main: paying for 22-input transactions creates a fingerprinted population, the card concedes it is farmable, and under uniform transactions it 'reduces to setting that constant low'. |
| Inflation bounty vault: a bug bounty adjudicated by arithmetic | By its own text it is blind to note forgery, grindable, and pays only once funds are stealable, and the public supply page is 'the free version of the same check'. Its useful piece, a funded quantum canary, moves into the supply audit card. |
| Instant-swap aggregator listing (desk as MM backend) | Duplicate: its own text says it overlaps the 'Desk as liquidity' card, which carries the operating plan, the venue facts and the inventory model. |
| ZKAS tipbot: non-custodial gift links in Discord, Telegram and X | Merged into the gift-links card: both cards said 'same plan, build them as one product'. The tipbot commands, the pre-registration trick and the caps now live there. |
| Deniable unanimity: the veto that leaves no trace | Merged into the baton-ledger card: the same unanimity-plus-expiry mechanism on the same blocked multi-party bundles, with the same flaw (any one member vetoes forever, unattributably). |
| Dark Tote: shielded pari-mutuel pool | Merged into the DLC card: the same unbuilt oracle and adaptor signatures, runnable only by a licensed operator. |
| Underwriter's key: crime cover priced on continuous proof | A tiny insurable population and years-long underwriting cycles; the mechanism is already in the selective-disclosure and completeness-key cards. |
/api/wallet/proof), unless the wallet turned on "private sends". Several cards said a default send was unprovable.Every file, line number, flag, constant and function a card cites was looked up in today's code, and every live figure re-measured. 40 more corrections on 38 cards, among them:
87a63c48, built and tested, not yet deployed.87a63c48); the explorer still has to be redeployed.The core map (consensus, security, cryptography, wallet sync, the wallet daemon, the node) had not been touched since 28 October. All 55 items were re-read against the code and the work of 29 Oct to 3 Nov: 22 rewritten, 13 added, 1 removed, and the plan tables updated. The main corrections:
d6e22ec5, the published releases and the live explorer, first on 28 October and again on 3 November 2026. Dev-month estimates are the specs' own and are unchecked. Details of open, exploitable issues are left out until the fixes are deployed. The idea map with every card is at zkas.info/plan-draft-public-v2.