Rebuild any DOM, exactly as the browser saw it.

Static HTML is a fraction of the real page. Reconstruction captures the tree after hydration, after interaction, and after every mutation the framework applied.

Request Access bolt View Pricing arrow_back Back to Home

water_drop

Hydration Capture

Captures are taken after client frameworks have mounted, so component output is present rather than the empty shell served first.

await hydrate() → stable tree

lock_open

Shadow Boundary Pierce

Open and closed shadow roots are traversed and re-attached into a single addressable tree with stable node paths.

pierce(closed: true)

layers

iframe Unrolling

Nested same-origin and cross-origin frames are unrolled with their provenance preserved for downstream rules.

frames.unroll() → depth 3

difference

State Vector Diffing

Successive captures are diffed structurally, so you see what changed rather than that something changed.

diff.structural() → 39 deltas

schema

Clean Tree Synthesis

The reconstructed tree is normalised — wrappers collapsed, presentation-only nodes dropped — before serialisation.

tree.normalize() → 12,840 nodes

replay

Replay Fidelity

Every node keeps the coordinates and selectors needed to re-target it on a later visit, so rules survive redesigns.

node.retarget() → stable

reconstruct.log

STREAMING

[00:00.412] hydrate       ok   framework=react@19
[00:00.688] pierce         ok   shadow_roots=42 closed=7
[00:01.104] unroll_frames  ok   depth=3
[00:01.290] normalize      ok   nodes=12840
[00:01.301] serialize      ok   bytes=418293
[00:01.302] done           confidence=0.9998