Make A2UI dataModelUpdate.valueArray first-class

[bkudiess]

Summary

The v0.8 dataModelUpdate.valueArray typed value was silently dropped by the native WinUI A2UI parser — DataModelEntry had no ValueArray field and ToJsonNode() returned null for it. Only valueString/valueNumber/valueBoolean/valueMap were handled.

This broke seeding an array into a surface's data model — most visibly a multi-select MultipleChoice bound to a path could never be pre-populated — even though valueArray is part of the v0.8 protocol (docs/a2ui/protocol.md §2.2 and docs/a2ui/data-and-actions.md).

What changed

• Parser (A2UIProtocol.cs): add DataModelEntry.ValueArray; ToJsonNode() emits a real JsonArray; new ParseValueArray/ParseArrayElement handle value-typed object elements ({"valueString":"x"}), bare-primitive tolerance (["a",1,true]), nested maps/arrays, and preserve a JSON null (or value-less {}) as a stable index slot so position-sensitive consumers don't see shifted indices.
• DoS guard (DataModelStore.cs): the existing 32-deep depth bound now recurses valueArray as well as valueMap, so deep nesting can't be smuggled in through an array.
• Security (SecretRedactor.cs): the JsonArray redaction branch now checks each element path against the registered/denylist set before recursing (mirroring the object branch). Previously arrays only recursed, so a secret seeded into an array (e.g. an obscured TextField bound to /codes/0) could leak through canvas.a2ui.dump.
• Docs (SKILL.md): correct the stale "arrays are not first-class" note.

Why it's safe

• Purely additive on the wire; existing scalar/valueMap behavior is unchanged and regression-tested.
• System.Text.Json TryGetValue<T> is strict (no cross-kind coercion), so the bare-primitive string→bool→number detection can't mis-tag a value.
• Malformed "both map and array set" deterministically prefers the map; never throws.
• The redaction change can only over-redact, never under-redact.

Tests

• A2UIDataModelArrayTests.cs (new) — parser coverage: strings, mixed scalars, maps, nested arrays, bare primitives, empty, and null/value-less slots, plus valueMap/scalar regressions.
• SecretRedactorTests.cs — array-redaction regressions: registered element, registered array parent, denylisted scalar inside an array.
• A2UIDataModelStoreTests.cs (new) — store base-path landing + valueArray depth-guard rejection + valueMap depth regression.
• A2UIControlMatrixTests.cs — end-to-end: MultipleChoice (multi) seeded via valueArray asserts both the snapshot array and the rendered ListView preselection (parser → store → renderer round-trip).

Validation

./build.ps1 (all projects), OpenClaw.Shared.Tests, OpenClaw.Tray.Tests, and the A2UI OpenClaw.Tray.UITests all pass (OPENCLAW_REPO_ROOT set per AGENTS.md).

Proof

I ran temporary proof-only harnesses locally to print the actual behavior, then removed those harnesses so they are not part of this PR.

Parser + redaction behavior

Input behavior exercised:

{"dataModelUpdate":{"surfaceId":"s","path":"/form","contents":[{"key":"picked","valueArray":[{"valueString":"red"},{"valueString":"blue"}]}]}}

Terminal output from the proof run:

valueArray path: /form
valueArray stored JSON: ["red","blue"]
registered secret paths: /codes/0
redacted dump JSON: {"codes":["[REDACTED]","5678"]}

This proves the parser stores a real JSON array at the scoped path and that a registered secret array element is redacted instead of leaking through dump/snapshot output.

WinUI rendered state behavior

The WinUI proof seeded a multi-select MultipleChoice with:

{"dataModelUpdate":{"surfaceId":"s","contents":[{"key":"picked","valueArray":[{"valueString":"g"},{"valueString":"b"}]}]}}

Terminal output from the proof run:

snapshot dataModel.picked JSON: ["g","b"]
rendered ListView selected tags: b,g

This proves the value flows through parser → data model store → surface snapshot → native WinUI renderer: the snapshot carries the JSON array and the rendered ListView has Green/Blue preselected.

Known limitations (pre-existing, out of scope)

• Nested array/map breadth is bounded only by the 1 MiB line cap — identical to the existing valueMap behavior; a shared element-count budget would be a separate change.
• BuildActionContext deep-clones a non-secret subtree without redacting nested secret-named descendants — pre-existing, applies to valueMap too, not introduced here.

[clawsweeper]

Codex review: needs maintainer review before merge. Reviewed June 25, 2026, 4:08 PM ET / 20:08 UTC.

Summary
The PR adds A2UI valueArray parsing/storage, array depth guarding, array-element secret redaction, updated A2UI skill docs, and parser/store/redaction/WinUI tests.

Reproducibility: yes. Source inspection of current main shows docs require valueArray while DataModelEntry, ParseEntry, and ToJsonNode omit it, so a dataModelUpdate with only valueArray stores null instead of a JsonArray.

Review metrics: 2 noteworthy metrics.

• Changed surface: 9 files changed, +498/-9. The patch is focused on A2UI parser/storage/redaction/docs/tests but broad enough to warrant normal maintainer review.
• Check rollup: 9 success, 6 skipped, 0 failing. The current GitHub check state supports merge-readiness review, though it does not replace maintainer approval.

Merge readiness
Overall: 🦞 diamond lobster
Proof: 🦞 diamond lobster
Patch quality: 🦞 diamond lobster
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Mantis proof suggestion
A short visible WinUI proof would help maintainers inspect the path-bound MultipleChoice preselection behavior directly. A maintainer can ask Mantis to capture proof by posting this exact PR comment:

@openclaw-mantis visual task: verify that WinUI A2UI MultipleChoice bound to /picked is preselected when dataModelUpdate seeds /picked with valueArray ["g","b"].

Next step before merge

• No automated repair is needed; with no concrete patch defects found, the remaining action is normal maintainer review or merge.

Security
Cleared: Cleared: the diff adds no dependencies, workflows, credential sources, or package-resolution changes, and the redaction update narrows an array-element leak path.

Review details

Best possible solution:

Land the focused protocol-parity fix after normal maintainer review, keeping valueArray handling in the existing A2UI parser/store/redaction path.

Do we have a high-confidence way to reproduce the issue?

Yes. Source inspection of current main shows docs require valueArray while DataModelEntry, ParseEntry, and ToJsonNode omit it, so a dataModelUpdate with only valueArray stores null instead of a JsonArray.

Is this the best way to solve the issue?

Yes. Extending the existing typed-value parser, store depth guard, redactor, and focused A2UI tests is the narrow maintainable fix for the documented protocol behavior.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 7a7f39df2397.

Label changes

Label justifications:

• P2: This fixes a documented A2UI parser/storage/redaction bug with limited blast radius to WinUI A2UI data-model updates.
• rating: 🦞 diamond lobster: Overall readiness is 🦞 diamond lobster; proof is 🦞 diamond lobster and patch quality is 🦞 diamond lobster.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): Sufficient: the PR body includes copied after-change live output showing parser/storage/redaction behavior and WinUI MultipleChoice preselection through the real A2UI path.
• proof: sufficient: Contributor real behavior proof is sufficient. Sufficient: the PR body includes copied after-change live output showing parser/storage/redaction behavior and WinUI MultipleChoice preselection through the real A2UI path.

Evidence reviewed

What I checked:

• AGENTS policy read: The repository AGENTS.md was read fully; its validation guidance was used as review context, but this cleanup review stayed read-only and did not run local build/test commands. (AGENTS.md:1, 7a7f39df2397)
• No maintainer notes found: No files were present under .agents/maintainer-notes, so there were no matching internal maintainer decisions to apply to these touched A2UI paths. (7a7f39df2397)
• Current docs require valueArray: Current main documents valueArray alongside scalar typed values and valueMap for dataModelUpdate.contents. (docs/a2ui/protocol.md:67, 7a7f39df2397)
• Current parser omits valueArray: Current main DataModelEntry exposes scalar fields and ValueMap, and ParseEntry only populates those fields, so a valueArray entry cannot become a stored JsonArray. (src/OpenClaw.Tray.WinUI/A2UI/Protocol/A2UIProtocol.cs:80, 7a7f39df2397)
• Current depth and redaction gaps: Current main bounds only nested ValueMap depth and the array redactor recurses without first redacting a registered scalar array-element path. (src/OpenClaw.Tray.WinUI/A2UI/Rendering/SecretRedactor.cs:104, 7a7f39df2397)
• Renderer already consumes arrays: The current WinUI MultipleChoice renderer reads path-bound multi-select selections from a JsonArray, so storing a real array is the narrow path to preselection. (src/OpenClaw.Tray.WinUI/A2UI/Rendering/Renderers/InteractiveRenderers.cs:420, 7a7f39df2397)

Likely related people:

• codemonkeychris: Authored the merged native WinUI A2UI renderer work that introduced the parser, data model store, secret redactor, and UI test surface touched by this PR. (role: feature introducer; confidence: high; commits: 9fa43f347703; files: src/OpenClaw.Tray.WinUI/A2UI/Protocol/A2UIProtocol.cs, src/OpenClaw.Tray.WinUI/A2UI/DataModel/DataModelStore.cs, src/OpenClaw.Tray.WinUI/A2UI/Rendering/SecretRedactor.cs)
• shanselman: Merged the native WinUI A2UI renderer PR and has adjacent test-safety history near this feature area. (role: reviewer and merger; confidence: medium; commits: 9fa43f347703, 394695cc3331; files: src/OpenClaw.Tray.WinUI/A2UI/Protocol/A2UIProtocol.cs, tests/OpenClaw.Tray.UITests/A2UIControlMatrixTests.cs)

What the crustacean ranks mean

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

How this review workflow works

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.