NIFT LABSAll experiments ↗

Experiment 03 / Capgo / October 2026

One website.
Two migration
philosophies.

What changes when we move a production-scale Astro website to Nift—and then reconsider its implementation for a world where agents do the routine editing?

1,347 routesThree finished implementationsOne pinned production reference

An independent experiment using Capgo’s public website. Not an official Capgo migration or a recreation of its private services.

01 / The premise

Same website.
Different questions.

We compared the original production Capgo / Astro build with two deliberately different Nift implementations. Agent is not simply faithful Capgo with more optimisation. The content and desired behaviour are the reference; the maintenance assumptions differ.

capgo / Faithful migration

How good can a faithful migration be?

Can we migrate a production-grade Astro site without throwing away its authored content model, routes, browser behaviour and existing runtime assumptions?

Fidelity comes first. Keep the existing mental model where practical, reconnect authored MDX/content and metadata, and migrate the build layer. Retain frontend implementation where that is the dependable route to parity.

Humans and agents, working together.

A conventional production codebase: humans engineer and review; agents assist with migration, content, validation and repetitive maintenance.

capgo-agent / Agent-native reconstruction

What if agents perform routine engineering?

Preserve the product, content and behaviour, while reconsidering the implementation around Nift, semantic HTML, CSS and maintained vanilla JavaScript.

Agents can navigate a large corpus, apply repetitive edits and validate many routes. Explicit mappings and deterministic primitives matter more than making every repeated fragment pleasant to hand-edit.

Agents implement. Humans direct and review.

Humans define intent, architecture, policy and acceptance. Source remains inspectable, debuggable, version controlled and reproducible; manual editing remains possible.

This tests a maintenance premise, not whether agents replace engineering judgement. Lower measured build cost does not prove lower long-term maintenance cost.

02 / Measurements

The finished builds,
on the same machine.

Full build time

Elapsed median · seconds · linear scale 0–110

Original Astro106.09 s
Faithful Nift5.67 s
Agent Nift2.23 s

Peak process memory

Maximum across runs · linear scale 0–4,000 MiB

Original Astro3.61 GiB
Faithful Nift235.8 MiB
Agent Nift101.7 MiB

Bar lengths start at zero and use the labelled linear scales. Memory bar lengths use raw KiB measurements converted to MiB; labels retain the table’s units. The table records sample counts and ranges.

Repeated full builds: elapsed medians and maximum peak process RSS across runs
ImplementationFull buildPeak process RSSRuns / elapsed range
Original Astro106.09 seconds3.61 GiB5 runs92.18–124.61 s
Faithful Nift5.67 seconds235.8 MiB3 runs5.56–5.67 s
Agent Nift2.23 seconds101.7 MiB3 runs2.21–2.34 s

Agent real source edit → normal rebuild1.04–1.09 sRange of the four measured edit-family medians; three runs per family.

Different work, measured honestly.

These are local measurements of three finished implementations, not equivalent-work compiler benchmarks. Astro renders authored content and regenerates frontend bundles. Faithful Nift retains compiled frontend assets and, in its warm build, reuses prepared MDX/content. Agent keeps already-converted HTML, retains production CSS/images/fonts, and independently bundles its maintained browser JS on every normal build. Agent has less content-conversion work to do.

Faithful edits change authored MDX/Markdown or marketing inputs; Agent edits change HTML. Cache states and authoring work differ. Private backend workflows remain uncertified in both Nift versions.

Derived from the final measurements: upstream-to-faithful elapsed ratio 18.7×; upstream-to-Agent 47.6×; faithful-to-Agent 2.54×. Upstream peak RSS is 15.7× faithful and 36.3× Agent. Ratios describe these workflows, not additional measurements.

Development iteration

What does an edit cost?

Iteration peak memory / RSS

Memory for changed-input production publications, separate from full-build memory and elapsed time. Linear bars start at zero; exact values are shown in MiB.

GNU time maximum individual process RSS; not aggregate simultaneous memory
0280 MiB

No-op

Faithful Nift235.4 MiB
Agent Nift101.8 MiB

Docs edit

Faithful Nift254.4 MiB
Agent Nift101.3 MiB

Rich docs edit

Faithful Nift275.6 MiB
Agent Nift101.2 MiB

Product / marketing edit

Faithful Nift236.3 MiB
Agent Nift101.5 MiB

Blog edit

Faithful Nift251.7 MiB
Agent Nift101.0 MiB

Maximum across three original runs; observed min–max shown. Existing targeted measurements retain this same maximum statistic. Full case tables and evidence below retain the other scopes and workloads.

Full builds tell part of the CI/deployment story. A maintenance loop is edit → build → validate → inspect → correct → repeat. The numbers below measure only build time until output exists; browser inspection, parity audits and human/agent review add time.

Faithful / ordinary real edits4.28–6.41 s

Authored MDX/Markdown or marketing inputs. Humans and agents share the content model.

Agent / ordinary real edits1.04–1.09 s

Mapped HTML fragments. Agents implement routine changes; humans direct and review.

Ranges above span the four edit-family medians, not all individual observations. Equivalent Astro edit and no-op loops were not measured; 106.09 s full-build time must not be substituted for its edit latency.

Ordinary nift build · elapsed medians · three runs each
WorkloadAstro edit loopFaithful NiftAgent Nift
No-opNot measured4.6 s1.08 s
Docs editNot measured5.78 s1.08 s
Rich docs editNot measured6.41 s1.06 s
Product / marketing editNot measured4.28 s1.09 s
Blog editNot measured4.35 s1.04 s

Explicit-target edit → rebuild

Agent-oriented workflow: an agent knows the page it changed and runs nift build <target>. Normal hooks still run. Across four families, faithful medians were 2.07–3.71 s and Agent medians 0.86–0.91 s.

Three temporary real edits per family · elapsed median (range) · maximum process RSS
Target / familyFaithfulAgentPeak RSS: faithful / Agent
Docs paragraph
docs/getting-started/quickstart/
3.21 s
3.06–3.39 s
0.91 s
0.85–0.94 s
255.9 / 101.6 MiB
Rich docs
docs/builder/getting-started/
3.71 s
3.65–3.74 s
0.86 s
0.83–0.91 s
275.3 / 101.5 MiB
Product title / text
native-build/
2.07 s
2.07–2.36 s
0.89 s
0.84–0.93 s
236.6 / 101.2 MiB
Blog article
blog/yarn-clear-cache/
2.33 s
1.87–2.38 s
0.87 s
0.86–0.89 s
236.4 / 101.6 MiB

All 24 edits reached output. Hash comparisons of served HTML/JS found exactly one changed HTML file per target and no changed JS files. Faithful hooks may still reconsider or prepare other content; preparation observations are retained in the raw evidence. Sources were restored after every sample. Agent ordinary restoration passed full parity; faithful needed a final unmeasured full rebuild to restore all output. Complete gate results are retained.

Targets select pages; they do not automatically rebuild all consumers of a shared change. Use an ordinary dependency-aware build or the complete affected target set for shared navigation, styles and controllers, then run relevant validation. Restoration caveat: faithful ordinary restoration sometimes reported all pages up to date while targeted test output remained. The final full rebuild restores it. Reverting a faithful target edit should explicitly rebuild that target or all output and verify it; the underlying cause is not established in this reporting checkpoint. Astro targeted/edit/HMR loops were not measured. Historical ordinary edit examples differ from this new suite; these are not paired ordinary-versus-targeted speedup measurements.

Shorter measured build waits reduce one part of repeated checking. They make frequent rebuilds practical, but do not establish total agent cost, review quality or productivity: those were not measured. Targeted builds matter when an agent knows the affected page; shared changes still require the complete affected set and broader validation.

Separate cold-content preparation measurements

Faithful cold content intermediates: 81.29 s median / 667.1 MiB maximum process RSS (three runs, 80.09–82.85 s). This removes prepared content/cache state, but retains installed dependencies and compiled frontend assets. Faithful web regeneration alone: 20.68 s / 667.6 MiB. Agent has no MDX preparation stage.

The historical single Agent 1.98 s observation is superseded. Faithful’s earlier 4.40 s checkpoint versus final 5.67 s is retained in its report; these unpaired samples do not establish the cause of that difference.

Iteration memory / peak RSS

Incremental memory matters alongside elapsed time: a smaller affected build can leave more headroom for concurrent builds, CI jobs and agents on constrained machines. Shared changes and retained worker pools can still use substantial memory; an incremental build is not guaranteed to have a lower peak than a full build. Actual concurrency and total workflow productivity were not benchmarked.

GNU time maximum individual process RSS; not aggregate simultaneous memory. Maximum across three original runs; observed min–max shown. Existing targeted measurements retain this same maximum statistic.

Peak RSS in MiB · accepted timings unchanged
CaseFaithful NiftAgent Nift
No-op235.4233.2–235.4101.8101.0–101.8
Docs edit254.4250.4–254.4101.3100.7–101.3
Rich docs edit275.6274.2–275.6101.296.4–101.2
Product / marketing edit236.3234.9–236.3101.5100.9–101.5
Blog edit251.7248.0–251.7101.096.0–101.0

Different scopes and hosts remain separate; full-build memory is not substituted for an edit. Raw measurements, provenance and memory scopes ↗

03 / The process

Start from known-good output.
Recover the structure beneath it.

  1. 01

    A golden reference

    Build the pinned official Capgo site successfully. Capture all HTML, CSS, JS, fonts, images and routes. That production output becomes the canonical oracle, avoiding a screenshot-inspired reinterpretation.

  2. 02

    Faithful → recover the authored site

    Begin with identical compiled output, then extract shared Nift templates, headers, footers and page structures. Reconnect MDX/Markdown, metadata and authored bodies while continuously testing output and route parity. The result goes beyond copying a compiled site into public/.

  3. 03

    Agent → reconsider the implementation

    Use official production Capgo as the visual/behavioural oracle and faithful Capgo as the content/route/source-mapping oracle. Maintain converted HTML and shared Nift structures; recover readable browser controllers and replace the normal Astro build with independent JS bundling.

  4. 04

    Validate, freeze, then measure

    Audit dependencies, test interactions, reproduce a fresh checkout and freeze architecture before repeated benchmarks. A real-edit check caught missing raw-fragment dependencies; explicit Nift declarations fixed propagation before accepting incremental measurements. No optimisation pass followed the final numbers.

04 / Corpus & confidence

A site, not a homepage.

1,347captured HTML routes3,876reference files in the output gate1,290source-connected authored bodies in faithful

Counted without double-counting

Authored docs bodies
529
Published blog bodies
514
Plugin tutorial bodies
149
Marketing/legal bodies
98
Source-backed listing layouts
51
Registry-only landing routes
2
Compatibility redirects
4

The first four rows total 1,290 authored bodies. With 51 listing layouts: 1,341 source-backed rendered regions. Docs include 519 MDX and ten Markdown inputs. Listing grids are not extra pages.

What the gates establish

Faithful regenerates 1,343 content/listing/registry heads; four redirect heads remain reference artifacts. All 1,347 parsed heads are byte identical. Its 1,192 accepted body-byte differences pass semantic equivalence.

Agent HTML is byte identical to the reference. Its gate checks retained assets, 21 rebuilt controller entries, 133 generated browser assets and the absence of 103 retired JS files. Both gates report zero introduced semantic/link issues and zero unexpected files.

The reference’s 11,339 existing link/anchor issues remain separately recorded, including missing translated destinations. We preserve upstream quirks rather than silently inventing fixes.

Browser behaviour is a separate check

Thirteen representative page classes at desktop and mobile cover homepage, pricing, product, solution, comparison, docs landing/normal/rich docs, blog listing/article, data, legal and a browser tool. Checks include navigation, dropdowns, tabs, copy buttons, search/filter, TOC, forms, pricing, languages and keyboard/focus behaviour. Actual Algolia search returned results; data-unavailable states matched production.

Of 26 paired viewport screenshots, 18 have no thresholded pixel difference; the remaining differences are below 1%, including animation timing. These are representative viewport checks—not exhaustive full-page pixel or accessibility certification.

What parity does not establish

Real account creation, authenticated workflows, device identifiers, genuine signing success and browser certificate upload/download remain uncertified. External content and availability may change. No accounts or external messages were created. Agent’s allowlisted compatible-provider adapter passes local mock tests; it does not recreate those services. Faithful’s original signing/device endpoints remain unsupported locally, and its older adapter does not automatically reconnect retained controllers.

Corrected Agent revision 6371356 passed an independent fresh clone with no copied project output, dependencies or Nift cache: locked install, first build, all routes, tracked inputs, fragment dependency checks and clean Git status. Installation used an existing npm download cache; this was not a cold-machine package-download test. The preceding bootstrap observation was 5.07 s / 165.4 MiB, excluding installation, and is separate from repeated warm builds.

05 / Maintenance

Who edits it matters.
So does what remains.

Three architectures, different maintenance strengths
ConcernOriginal AstroFaithful capgocapgo-agent
Content modelMDX/frontmatter, components, StarlightExisting authored corpus, explicit render adaptersConverted HTML fragments, shared Nift includes
Routine editingEstablished product engineering workflowHumans + agents; familiar content authoringAgents implement; humans direct and review
Browser ownershipUpstream components and build pipelineRetained compiled frontend implementationReadable recovered controllers, separately bundled
Normal buildRender content, compile frontend assetsNift + cached content preparation; retain frontend bundlesNift + JS bundling; HTML remains converted
Best maintenance fitExisting collection/compiler conveniences and deployed worker systemOngoing Nift migration preserving MDX authoringExplicit agent-mediated HTML/vanilla workflow
Cost to considerMeasured build time and larger build surfaceAdapters and retained assets must stay coordinatedMapped bulk edits; retained CSS needs an explicit change pipeline

Simpler build ≠ less shipped JS

Agent has no normal Astro build/runtime requirement or React install, and no Astro hydration islands. DocSearch is a documented Preact compatibility exception for modal, keyboard, focus, ranking and history state. Mermaid, SDK, auth and crypto libraries remain. Production inline controllers and external integrations are explicitly retained, not claimed newly invented vanilla code.

Total rebuilt Agent JS is 6,082,284 bytes, compared with original 6,073,203 bytes: slightly larger. There is no demonstrated overall client-JS reduction. These are uncompressed asset totals, not per-page downloads. All versions retain production CSS/images/fonts.

Inspectable, but substantial

Agent maintains 1,347 wrappers, 1,881 shared HTML includes, 5,990 converted fragments and six templates/helpers. Thirty-seven readable JS files supply all 21 entry controllers. Its locked dependency set has 187 packages and 228,622,975 bytes of installed regular files.

Faithful’s root lock has 188 packages, plus the separate MDX renderer and other inputs; upstream’s Bun monorepo scope differs. Equivalent total install/download footprints were not measured, so no percentage saving is claimed. Agent keeps 19 original CSS files and removes only audited, unreferenced JS from served output.

What would we maintain?

If we were the existing Capgo organisation today

Staying with upstream avoids migration and new certification work, and retains its deployed translation/service system. That is an organisational-risk argument, not a demonstrated development-loop advantage. Team familiarity is unknown, private workflows remain uncertified, and comparable Astro dev/HMR/edit measurements are missing.

If we were committed to Nift, with humans + agents

We would prefer faithful when retaining MDX/content authoring matters. Its captured corpus, source-backed metadata and shared composition are demonstrated; ordinary real-edit family medians are 4.28–6.41 s, with explicit targets at 2.07–3.71 s here. Schema/discovery and new style/asset-generation conveniences still need acceptance or further tooling.

If we were designing for agent-mediated implementation

We would prefer Agent for that premise: explicit source mappings, inspectable markup, readable controllers and parity gates, with normal real-edit family medians of 1.04–1.09 s and targets at 0.86–0.91 s. Those timings support cheap build feedback; they do not measure total validation latency, agent cost or editing reliability. Humans still direct/review, and shared changes require broader builds and checks.

A preference for a Nift engineering workflow and caution about a real production migration can coexist. We cannot decide developer experience from the Astro full build alone.

If migration weren’t part of the equation

Assume all three implementations already exist, are equally production-certified, have equivalent access to the required backends, and the team can work effectively with either stack. Migration, switching cost and incumbency are excluded. The following are engineering preferences based on this corpus, not measured long-term productivity results.

Humans + agents: faithful Nift

We would choose faithful Nift. It keeps authored MDX/Markdown, frontmatter and shared composition readable for people while giving agents explicit mappings and parity gates for corpus-wide work. Measured ordinary real-edit family medians are 4.28–6.41 s and targeted medians 2.07–3.71 s. We prefer that combination of familiar content authoring, programmable preparation and a small measured build budget over maintaining converted HTML by hand.

Astro still provides more complete collection validation/discovery, typed component and scoped-style compilation, and packaged build adapters. Faithful’s bounded adapters and explicit route registration need care; retained compiled browser assets make new browser behaviour and CSS changes less direct to debug and regenerate. Those are real maintenance costs we would accept for this workflow, not capabilities we claim to have finished. Equal backend access removes the worker deployment advantage; translation remains a separate system in either case. Dependency totals are not comparable across the installs, and retained browser libraries remain substantial. Astro edit/HMR was not measured, so this preference does not claim that Nift beats Astro’s day-to-day latency.

Agent-primary: Agent Nift

We would choose Agent Nift. Deterministic source mappings, shared includes and readable controller ownership make routine edits and debugging explicit. Agents can make mapped bulk changes while humans review intent, diffs and acceptance. Its measured ordinary real-edit medians of 1.04–1.09 s and targeted medians of 0.86–0.91 s support repeated build feedback. Independent JS bundling and the absence of a normal Astro build or hydration islands simplify the build architecture; route, dependency, parity and representative browser gates supply concrete checks.

This choice accepts a large fragment corpus and an explicit workflow for new routes, styles and shared changes. It does not imply less shipped JavaScript, fewer total dependencies, exhaustive validation, or proven lower agent cost. Browser SDKs and the DocSearch Preact exception remain. Targets alone do not cover all consumers of shared changes; broader builds and browser checks remain necessary. The documented faithful restoration issue also shows why parity checks belong in the maintenance loop. With incumbency neutralised, we prefer Agent’s inspectable HTML and controller pipeline for agent-led implementation, and faithful’s authored content model for human + agent work.

Maintenance means changing things

Source-traced workflows; shared-change and browser-test timings remain unmeasured
ChangeOriginal AstroFaithful NiftAgent Nift
Docs paragraphMDX; dev/HMR latency not measuredAuthored MDX; measured ordinary/targeted buildsMapped HTML; measured ordinary/targeted builds
Shared navigationHeader/menu or sidebar configShared shell or recovered configShared include / parameterised shell
Shared visual componentComponent/style compiler pipelineRegistered adapter; CSS generation separateShared HTML/CSS or mapped bulk edits
Browser behaviourClient source + bundlerRetained bundle needs source recovery/rebundleReadable frontend/ sources; hook bundles all entries
LocaleDefinitions + separate translation workerEnglish head/menu mappings + worker deploymentEnglish HTML/menu mappings + worker deployment
Hundreds of pagesCollection queries or scripted source editsAuthored corpus + dependency graphExplicit deterministic mappings + agent bulk edits

For shared changes, build the complete affected set and inspect relevant desktop/mobile behaviour. Intentional new content/design requires updating the golden policy, not treating old byte equality as acceptance. We did not measure source-location counts or human/agent effort for these scenarios.

Content: a concrete convenience, not a capability verdict

Upstream’s glob-loaded blog collection declares Zod fields, supported locales, dates and tag normalisation; routes query the collection. Adding a published source entry is discovered by that pipeline. Faithful retains authored MDX/frontmatter and discovers English blog files for listings, but new rendered routes also need body/metadata mappings and Nift registration. Its narrow YAML adapters do not apply the complete upstream collection schema. Agent edits mapped converted HTML and explicitly registers routes.

These are current migration/tooling gaps and different workflows. Nift already composes authored sources, MDX, data, templates and dependencies; a schema validator or route generator can run in preparation. No fundamental content-modelling limitation was demonstrated.

Composition: compare the operation

Capgo uses typed prop declarations, named slots and compiler-scoped styles. Faithful’s registered component adapter already handles parameters, children and named slots; Agent uses shared Nift includes/functions and readable browser controllers. Exact rendered output demonstrates sufficient composition for this corpus. Arbitrary new Astro syntax or freshly scoped CSS is not guaranteed by the bounded migration adapter. Upstream packages that compiler workflow; this is a concrete convenience difference, not proof that Nift cannot compose components.

Localisation: the translation worker does the work

The shared Astro configuration builds English only. Nine advertised language choices point to a separate Cloudflare edge worker, which translates the English origin, stores page/state records, rewrites navigation, language metadata and hreflang, and implements pending-translation retries/fallback. Shared copy is English with generated translation contexts. The Nift versions preserve English output and language links; neither deploys or certifies that worker.

Adding a locale requires changes to web/docs locale definitions and the worker’s supported set/configuration, then deployed validation. We did not measure comparative file counts or effort. The existing worker is an upstream implementation-coverage advantage; the claim that Astro’s i18n model is inherently better for Capgo is withdrawn.

Integrations: turnkey adapters versus programmable seams

Astro’s configured MDX/Starlight, DocSearch, sitemap, icon and Tailwind/Vite packages offer concrete ready-made conveniences. Nift does not require a dedicated framework plugin for each underlying system. Faithful’s hook runs Node content, code and metadata preparation, processes recorded API JSON and consumes generated files with dependency tracking. Agent’s hook independently bundles browser packages with esbuild; its allowlisted provider server composes compatible HTTP services. These are demonstrated orchestration seams.

Algolia, Mermaid, Supabase, metrics, pricing, embeds and browser SDKs connect through ordinary packages, configuration, HTTP or browser code. Translation and signing/device services require separate deployed backends. Retained sitemap/LLM/raw-Markdown assets are not proof of automatic freshness after new content. No inspected integration established an intrinsic inability of Nift to connect; several ports and live workflows remain incomplete. Native/FFI was not exercised and is not counted as evidence here.

Available adapters are distinct from general interoperability. Community size, documentation quality, team familiarity and long-term editing productivity were not quantified. Every identified Nift drawback is classified as an experiment coverage gap, migration tooling gap or different workflow in the source-backed maintenance investigation. It inventories integration families and traces concrete editing operations; no implementation was changed for this reporting correction.

06 / Methodology

Enough detail to disagree.

Machine, commands, revisions and measurement
Machine
Intel i7-12700H, 20 logical CPUs
System
Ubuntu 26.04.1 LTS; Linux 7.0.0-29
Nift
4.6.0; build threads −1 (all available cores)
Tooling
Nift: Node 22.22.1. Upstream: Bun 1.4.2 / Node 24.21.0
Source references
Upstream 7d5b69d6; faithful frozen 590c1eec; Agent final dd95a37, measured 6371356
MDX renderer
Faithful package revision 22bb53653f47c22defff863e3465e8df9a11da99. Agent does not use MDX conversion.
# Original: 5 runs
bun run build

# Each Nift implementation: 3 full runs
nift build --all

# No-op and each real-edit family: 3 runs
nift build

# GNU time records elapsed and maximum process RSS
/usr/bin/time -v -o measurement.time nift build --all

The explicit-target suite used three temporary real edits per family, warm preparation, normal hooks and nift build <target>. Exact sources, targets, hashes, output fanout and restoration gates are recorded in the targeted evidence at b555abe. Median means the middle elapsed observation in each repeated set. Peak RSS is the maximum GNU time process RSS across that set, not aggregate simultaneous process memory. Runs were serial, dependency installation excluded, and OS page cache not flushed. Nift warm builds followed an unmeasured prime. Real edits changed visible source text, verified propagation to output, then restored source/output outside the measurement.

Astro’s command clears its per-app output/Astro/deploy caches each run while retaining dependencies, committed data and recorded plans/credits/rankings public-response preload. Other optional fetches retain upstream behaviour. Public snapshots are explicit inputs; a credentialless empty-live-sponsor fixture is labelled, not invented success. Reference dates/randomness are recorded.

Faithful warm runs retain content/MDX and output caches plus compiled frontend assets. Agent retains converted HTML, assets and ordinary Nift hash state; its native esbuild hook runs each time. The raw records document cold-intermediate and web-regeneration classes separately.

The installed Nift executable does not expose its source commit. Its exact identity is SHA-256 03ed5ed8344280c736e78a18f4ad80c6e92e3f45e63ff3784523aa122f0ac0b6; raw methodology includes lock hashes and dependency pins. We do not substitute an assumed source revision.

07 / Open notebook

Follow the evidence.

08 / What we learned

Migration is a process.
Maintenance is a choice.

Nift supported a production-scale content migration with high fidelity, without discarding the authoring model. Reconsidering that model around converted HTML and independently maintained controllers reduced the measured build work further.

The benefit is not universal: retained frontend assets, preparation caches, library dependencies and private-service gaps all matter. Agent-mediated editing makes repetitive source representations plausible; it does not remove the need for review, debugging or product decisions.

This work informs future migration scaffolding, source mappings, parity gates and handover support—perhaps a command such as nift init --migration. That is a proposed direction, not an available command.

Return to the experiment notebook →