Ascension cycle implementation handoff — historical snapshot
This records the October 14, 2026 implementation handoff, when the candidate was unpublished. Its source identities, results and remaining requirements are retained as historical evidence. For the subsequent combined release, use release readiness and the 1.31.0 release notes; do not treat the draft states below as a live PR status. The complete-cycle demonstration and focused component results are separate from the 80 requirement decisions, broad repository gates and operational Mainnet commissioning. A passing fixture does not authorize a production deployment or establish realized commercial value.
Source and publication identity at capture
| Item | Recorded identity / state |
|---|---|
| Inspected Nova-Seed source | 89b6f78b83a6643416649173d7d16b6d1c7a65c1, existing PR #4779 |
| Merged MARK base | c38f430d5a07064ad7c19ae761e91eaad3e48efe, PR #4778 |
| Native runtime draft PR | #4780 |
| Native source head | 98ef5a5d510b81048ed32cad06e132d30b166e8f; includes encoded request-size boundary correction |
| Hosted source draft PR | #4782, based on native #4780 |
| Hosted source head | 9f3b6b84ce7ff242b33a7dbd8d2cac34f32289b7; documentation/evidence delivery commits follow without changing these executable sources |
| Release tag | None created for this implementation |
| Verified published frontend/version | Not published |
| Production write qualification | No added qualified production profile; original archived restrictions retained |
The reports retain exact executable-source hashes, fixture-only profile changes, commands and dependency versions. The coordinator independently matched all 34 changed production/test files to these commits. At this capture, both PRs were unmerged drafts. A later source change invalidates applicability to that changed candidate until the affected check is rerun. Imported artifacts and a matching chain ID do not establish an exact deployment or current authority.
What changed and why
| Existing integration point | Necessary change | Preserved boundary |
|---|---|---|
docs/ascension/mainnet/index.html, Studio app/workflow/styles |
Extend the existing journey through exact native work, evidence, accounting, evaluation and a distinct successor | Same route, wallet client, session, original journal, atomic lock and explicit transaction confirmation |
| Studio package/storage helpers | Bounded signed/native and private-project records; publication of exact result bytes | Existing package/capsule/hash domains and readable recovery semantics remain explicit |
Native SUCCESSOR CLI and runtime/api.py |
Add additive cycle commands and authenticated local operator forms/actions | Existing runtime trust, bearer/Origin/Host policy, signed journals and action-time grants |
| Native operator navigation | Link the installed /ascension-studio desk and typed first installation |
No work-import key creation, generic config overwrite or hosted unauthenticated localhost access |
| MARK evidence workflow/inventory and contract harness | Include new executable assets, source hashes and genuine native/browser/EVM checks | Disposable-network guard and production empty registry remain enforced |
| Existing Nova-Seed guide | One link to the complete-cycle guide | Previous first-cycle instructions remain available |
New modules hold project history/private recovery, native request/result verification, rights-reviewed artifact reuse, receipt-reverified historical settlement flows and nonbinding owned-fund proposals. The normal browser/native journey uses forms and portable files. An administrator still installs trusted identities and optional private signer/role-token slots before first use.
The original BACKGROUND graph is byte-identical to the inspected source, with twelve nodes and eighteen
directed edges. Its SHA-256 is 5185e8312dbfa915915c9270dc806e7f47068f581adbb93344699672eff91331.
The engineering guide maps each direction to schemas/APIs, actor permission, preconditions, evidence,
recovery and acceptance IDs. At this capture, the coordinator reported an empty protected diff for original Solidity,
deployments, BACKGROUND, wallet client/session, service worker and dependency locks. Final diff review
of a later integration must separately account for intended shared-wallet hardening and exclude unintended
generated model, worker, WASM and binary outputs. This historical empty diff is not the later release's
change inventory. These checks do not replace full preservation or broad repository regression results.
Evidence and requirement decisions
The machine-readable acceptance index retains all original 80 IDs and their exact requirement texts. The index records 10 passed, 65 partial and 5 blocked. These are requirement decisions, not unit-test counts. Software completion and operational Mainnet completion remain blocked.
| Check | Recorded scope / result at capture |
|---|---|
| Project model | Final source-stable report: 14/14 passed, Node 22.17.1; encrypted recovery and global evaluation-ID uniqueness included |
| Accepted artifact reuse | Final source-stable report: 4/4 passed, Node 22.17.1; real numerical workbench with a synthetic chain verifier |
| Receipt ledger/proposals | Final source-stable report: 4/4 passed, Node 22.17.1; native-format signatures/ABI events with synthetic read-only RPC |
| Native cycle/Ascension/workbench/jobs | 59/59 passed; genuine fixed computation, synthetic RPC; exact source/dependency hashes retained |
| Installed local operator/API/browser | 28/28 passed; actual typed setup/pairing/separate-role forms and UTF-8 512 KiB request boundary; administrator trust/key/token provisioning remains explicit |
| Browser/native result verifier | 12/12 passed; exact original signatures, verdict/protocol integrity and cross-cycle evaluation attribution |
| Complete Studio/native/disposable EVM | Final corrected-source production-browser run passed: three genuine jobs, two cycles, 33 axe audits with zero violations; all 46 recorded dependencies matched |
| All Studio Node tests | 48/48 passed on pinned Node 22.17.1 and qualified Python |
| Existing MARK / contracts | 117 Node + 4 Python MARK tests and 108 contract tests passed; final built MARK browser also passed |
| Broad repository tests | Failed: 2,191 passed, 38 failed, 122 errors, 94 skipped, 4 xfailed; 80.57% coverage. The gate remains environment-contaminated; coverage alone does not pass it |
| Full-repository mypy | Failed: 222 diagnostics in 53 files versus exact-base 208 in 52 files; 14 new API/operator decorator errors, no removed baseline errors |
| Lint and documentation build | Earlier all-files run passed all 18 hooks. Final 116-path check passed 16 hooks with 2 no-files skips; strict assembled MkDocs passed in 1.92 seconds |
| Remote MARK acceptance CI | Run 38513745974 passed at hosted source 9f3b6b84ce7ff242b33a7dbd8d2cac34f32289b7; all 23 reported steps succeeded and three artifacts were uploaded |
The complete-cycle report demonstrates the paid accurate
negative report, separately frozen fresh numerical comparison, reviewed version choice, actual accepted
recipe consumption on new inputs, new
assignment/run/result/review/settlement, receipt-rechecked historical flows, explicit owner funding and
fresh-session encrypted recovery without sends or execution. Its actual browser report checks exact
topology before work and after restore. The example bundle preserves original signed bytes, public-safe fixture artifacts,
encrypted recovery and screenshot digests; it excludes private homes/keys/tokens. The
repository example is text-only. The separately saved public archive
contains both inspected original screenshots and has SHA-256
697ec0bd6e20ac57eea9a4e69a90148da0281658400e0f99802f3d20cc69772d (3,764,272 bytes).
Public fixture passphrases and local decrypt/validation instructions are included.
The evidence index retains commands, environments, source hashes and
scoped outcomes. The earlier Nova regression remains labeled as its prior integration snapshot; final
remote CI reran both Nova and the complete cycle at the actual hosted source commit. No report is edited
to pretend that its original execution happened at a later commit. A disk-full incident during final
documentation and broad-suite work was cleared by removing reproducible caches and an obsolete setup
environment. The broad-suite owner identified disk/temp-file evidence in 29 failures and all 122
errors. All 151 affected outcomes then passed in isolated or bounded reruns with the same qualified
environment. Seven of the nine other failures passed on retry. Two notebook kernel-start failures
reproduced at the unchanged inspected base with libzmq Operation not permitted. The 158 follow-up
passes do not convert the original failed run into a clean whole-suite pass.
Computer Use unit checks recorded 15 passed and one actual-container skip; this supports partial
recovery coverage, not live model/container qualification. There is no broad pytest success claim.
A separate saved diagnostic archive, ascension-regression-diagnostics.zip, retains the original
full-run log, JUnit, coverage, environment freeze and every bounded/baseline follow-up, including the
initial missing-assets failure. It contains 29 files, is 277,956 bytes and has SHA-256
cbe6ecb7eab48841e4de17afeb906dece23834ebcd049b623590d0a007618185. It is separate from the
public synthetic example; its original traces are not duplicated as repository binary files.
Supported and unverified environments
| Area | Implemented / exercised scope | Remaining qualification |
|---|---|---|
| Executor | Installed bounded native numerical work in the complete cycle; public-data and numerical Studio forms | Other native work families retain their documented scope; no general-agent or paid-model claim |
| Host | Linux, Python 3.12.14 and Node 22.17.1 in retained final runs | No new Windows/macOS or every-supported-Python-version trial claimed |
| Browser | Actual production Studio and installed operator in Chromium; EN/FR and declared responsive sizes | Human screen reader, physical mobile and low-resource trials remain separate |
| Wallet | Existing explicit injected-provider selection and exact EIP-1193 fixture against actual disposable EVM | No physical extension/mobile/hardware compatibility trial; no added WalletConnect or smart-account support |
| Storage | Existing supported HTTP adapter and exact-byte retrieval against local fixture storage | Actual authorized service-account/public-network propagation remains unperformed |
| Provider / Computer Use | Existing bounded Computer Use remains separately documented | No live paid-provider/model execution, useful commercial work or cost-saving claim |
| Review / rights | Explicit local roles, frozen findings and operator/rights-holder declaration | Fixture keys are not independent beneficial owners; no independent legal/security audit |
Deployment policy
| Deployment | Permitted scope |
|---|---|
| October 11 Mainnet archive | Exact historical inspection and contract-permitted recovery/obligations; new exposure remains blocked |
| Disposable exact-source fixture | Actual current-source EVM actions with mock external dependencies, synthetic assets and declared fixture actors only |
| New production deployment | Blocked until exact source/dependency/authority qualification, authorized canary and separate activation decision |
Frontend rollback does not reverse a chain transaction, pause immutable contracts, revoke a native grant or erase obligations. Preserve legitimate exits and pending-intent recovery while suspending unsafe new exposure through the existing policy boundary.
Review findings and remaining blockers
Integration review exercised hashing/signatures, original-byte recovery, uint64 timestamp preservation, controller/acceptance separation, role-token slots, preposting terms, action-time revocation, imported verdict/protocol edits and cross-cycle evaluation attribution. Final source regressions retain the SHA-256/Keccak distinction, lossless signed timestamps, exact UTF-8 POST limits and rejection of recovered cross-cycle evaluation attribution. This is project review with adversarial tests, not an independent external security audit.
| Remaining item | Owner | Objective clearance |
|---|---|---|
| Broad software/release gates | Integration owner and repository maintainers | Exact final-source lint/build/preservation/tests/coverage and required CI results; baseline failures reproduced and resolved, not waived |
| Full-repository type checking | Native/API owners and repository maintainers | Resolve the 14 added API/operator decorator diagnostics and retain a reviewed final-source full typecheck; scoped owned-file checks do not clear this gate |
| Complete accounting scope | Accounting/integration owner | Retained principal-return receipts and every claimed capital movement/branch; historical settlement view must not be represented as a complete wallet ledger |
| Shared project/device allocation | Accounting/integration owner | A supported shared reservation/allocation mechanism with adversarial concurrency evidence if allocation is claimed; present proposals remain nonbinding and reserve nothing |
| Per-edge observed state/actions | Product/integration owner | Each required graph direction exposes its actual evidence and next permitted action; topology alone does not close this requirement |
| Interruption/deadline/scale drills | Runtime/wallet/operations owners | Each named case and recovery point has retained current-source evidence and an accountable closed-browser handoff |
| Actual storage qualification | Authorized storage-account operator | Bounded real upload and independently retrieved matching public bytes with sanitized receipts and cost/retention scope |
| Human accessibility / wallet devices | Human testers and wallet compatibility owners | Actual declared browser/assistive-tool/device versions, consent/error/recovery results and limitations |
| Live provider trial | Authorized provider/runtime operator | Exact model/provider, bounded real execution, observed usage/costs/actions and reviewed output |
| Independent real actors / Mainnet canary | Deployment governance and consenting role holders | Exact qualified deployment, independent control evidence, bounded authorized lifecycle and distinct live continuation, reconciled receipts/obligations |
| Publication | Release maintainer | Reviewed final source, immutable release identity and independently verified served routes/assets; public activation remains a separate decision |
This capture did not establish a publishable integrated release. Later release decisions must identify the actual combined source and new evidence, retain the limits above, and distinguish software publication from external-service or Mainnet commissioning. The complete operational guide is ASCENSION_CYCLE.md and the graph/schema/authority contract is ASCENSION_CYCLE_ENGINEERING.md.