Traversing The Semantics Tree logo

Traversing The Semantics Tree

Community
skydoves
traversing-the-semantics-tree

Use this skill to navigate from one Compose semantics node to its relatives via `onParent`, `onChildren`, `onChild`, `onChildAt`, `onSibling`, `onSiblings`, `onAncestors`, plus the collection helpers `onFirst`, `onLast`, `filter`, `filterToOne`, and the `[index]` operator. Covers when to traverse vs when to add a stable `testTag`, the LazyColumn/LazyRow caveat (only currently composed children appear), the absence of a singular `onAncestor`, and the sticky `useUnmergedTree` flag across navigation. Use when the developer mentions `onChildren`, `onChild`, `onParent`, `onSiblings`, `onAncestors`, `filterToOne`, `onFirst`, `onLast`, brittle child-index chains, or asks how to find the second child of a Row, the parent of a Text, or any sibling of a node. If the developer is dot-chaining navigation through a layout, use this skill.

Overview

Publisherskydoves
Repositoryandroid-testing-skills
Skill nametraversing-the-semantics-tree
Stars
338
Forks
14
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by skydoves on GitHub. Read the source before you install it.

Installation

Install the Traversing The Semantics Tree AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/skydoves/android-testing-skills/tree/main/compose/finders/traversing-the-semantics-tree
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/skydoves/android-testing-skills.git /tmp/android-testing-skills
mkdir -p .claude/skills
cp -r /tmp/android-testing-skills/compose/finders/traversing-the-semantics-tree .claude/skills/traversing-the-semantics-tree
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Traversing The Semantics Tree in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Traversing The Semantics Tree on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Traversing The Semantics Tree is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Traversing the Semantics Tree — When a Single Finder Won't Reach the Node

The single-finder shortcuts (onNodeWithTag, etc.) cover ~95% of test queries. The remainder need tree navigation: "the parent of this Text", "the third child of this Row", "any sibling that is enabled". This skill maps those navigation operators, calls out the LazyColumn snapshot caveat, and shows when traversal is the wrong tool.

When to use this skill

  • The target node has no stable testTag and one cannot be added to production (third-party composable, dynamically generated children).
  • The test verifies structural relationships ("the second child of the row is the icon").
  • The developer chains .onChildren()[i].onChildAt(j) and wants to know whether that is the right shape.
  • The developer mentions onChildren, onChild, onParent, onSibling, onSiblings, onAncestors, filterToOne, onFirst, onLast, or [index].
  • A LazyColumn test misses items because they are off-screen.

When NOT to use this skill

  • A Modifier.testTag(...) could be added to the target node — adding a tag is almost always cleaner than a traversal chain. See ../finding-nodes-by-tag-text-content/SKILL.md.
  • The relationship is "anywhere above" / "anywhere below" — prefer the matcher-based hasAnyAncestor / hasAnyDescendant (see ../composing-semantics-matchers/SKILL.md).
  • The query is about a LazyColumn item by key — use performScrollToKey instead (see ../../actions/clicking-and-scrolling/SKILL.md, ../../patterns/testing-lazy-lists/SKILL.md).

Prerequisites

  • A working ComposeTestRule / ComposeUiTest. See ../../setup/configuring-test-dependencies/SKILL.md.
  • Familiarity with the merged vs unmerged tree distinction. See ../finding-nodes-by-tag-text-content/SKILL.md.

Workflow

  • 1. Pick the navigator by relationship type. Each operator returns either a SemanticsNodeInteraction (single — fails on 0 or >1) or a SemanticsNodeInteractionCollection (plural — never fails on 0).

    From a single node, go to …APIReturnsFile:line
    parentonParent()singleSelectors.kt:36-42
    exactly one childonChild()singleSelectors.kt:71-77
    child at an indexonChildAt(index)singleSelectors.kt:85
    all currently-composed childrenonChildren()collectionSelectors.kt:53-59
    exactly one siblingonSibling()singleSelectors.kt:119-125
    all siblingsonSiblings()collectionSelectors.kt:101-107
    every ancestor up to rootonAncestors()collectionSelectors.kt:140-146
    From a collection, narrow to …APIReturnsFile:line
    firstonFirst() (= [0])singleSelectors.kt:156-158
    lastonLast()singleSelectors.kt:168-170
    nth[index]singleSemanticsNodeInteraction.kt:261-267
    filter to a sub-collectionfilter(matcher)collectionSelectors.kt:178-186
    filter to exactly onefilterToOne(matcher)singleSelectors.kt:198-206
  • 2. Do not look for onAncestor (singular) — it does not exist. The API surface ships only onAncestors() plural (Selectors.kt:140-146). It returns [parent, grandparent, …, root] in that order. To assert "the immediate parent satisfies X", use onParent().assert(matcherX) or the hasParent(matcherX) predicate.

  • 3. Remember the useUnmergedTree flag is sticky. Every navigator carries the same useUnmergedTree value as the source SemanticsNodeInteraction (SemanticsNodeInteraction.kt:42-49; each Selectors.kt constructor passes useUnmergedTree through unchanged). A finder created with onNodeWithTag(tag, useUnmergedTree = true) keeps the flag on across onChild/onParent/filter. Skydoves hot take #2: default merged, flip to unmerged only when targeting composition detail.

  • 4. onChildren() is a snapshot at invocation time. It returns nodes "currently present in the semantic tree" (Selectors.kt:46-52). For a LazyColumn or LazyRow only the on-screen items appear. To reach an off-screen item, scroll first with performScrollToIndex / performScrollToKey / performScrollToNode (see ../../actions/clicking-and-scrolling/SKILL.md).

  • 5. Re-finding by tag usually beats navigating. If the tree shape might change between Compose versions or under translation rotation, a stable tag on the target node is more durable than a [2].onChildAt(0) chain. Add the tag in production. Skydoves hot take #1.

  • 6. Use filterToOne(matcher) when the count assertion is implicit. filterToOne throws on 0 or >1 matches at fetch time, the same way onNode(matcher) does (Selectors.kt:198-206). It is the collection-narrowing analogue of onNode(matcher).

Patterns

Pattern: prefer a tag over a deep [index] chain

kotlin
// WRONG
@Test
fun second_avatar_isVisible() {
    rule.setContent { ProfileGrid(profiles = profiles) }
    rule.onNodeWithTag("ProfileGrid")
        .onChildren()[2]
        .onChildren()[0]
        .assertIsDisplayed()
}
// WRONG because: layout shuffling (a header inserted, an extra wrapper, a Spacer added)
// silently changes which node is being asserted. The test passes for the wrong reason.
kotlin
// RIGHT
// production:
@Composable
fun ProfileGrid(profiles: List<Profile>) {
    LazyColumn(modifier = Modifier.testTag(ProfileGridTag)) {
        itemsIndexed(profiles) { index, profile ->
            Row(modifier = Modifier.testTag("$AvatarTagPrefix$index")) {
                Avatar(profile, modifier = Modifier.testTag("$AvatarImageTagPrefix$index"))
            }
        }
    }
}

// test:
rule.onNodeWithTag("$AvatarImageTagPrefix${1}").assertIsDisplayed()

Pattern: traversal when no tag is available

kotlin
@Test
fun row_third_child_isIcon() {
    rule.setContent {
        Row(modifier = Modifier.testTag("toolbar")) {
            Text("Title")
            Spacer(Modifier.weight(1f))
            Icon(Icons.Default.Share, contentDescription = "Share")
        }
    }

    rule.onNodeWithTag("toolbar")
        .onChildren()
        .assertCountEquals(3)

    rule.onNodeWithTag("toolbar")
        .onChildAt(2)
        .assertContentDescriptionEquals("Share")
}

onChildAt(index) is exactly onChildren()[index] (Selectors.kt:85). Both fail if the index is out of range or if the resolved node count is not exactly 1 at the leaf.

Pattern: filterToOne instead of [i]

kotlin
// WRONG
rule.onAllNodesWithTag(RowTag).onChildren().filter(hasClickAction())[0]
    .assertHasClickAction()
// WRONG because: indexing into a filtered collection silently passes when the filter returns
// many. The test asserts only "at least one is clickable", not "exactly one".
kotlin
// RIGHT
rule.onAllNodesWithTag(RowTag).onChildren()
    .filterToOne(hasClickAction())
    .assertHasClickAction()

Pattern: assert the parent role from a known child

kotlin
@Test
fun submitText_isInsideEnabledButton() {
    rule.setContent {
        Button(onClick = {}, modifier = Modifier.testTag("submit"), enabled = true) {
            Text("Submit", modifier = Modifier.testTag("submitLabel"))
        }
    }

    rule.onNodeWithTag("submitLabel", useUnmergedTree = true)
        .onParent()                       // sticky: stays unmerged
        .assertIsEnabled()
        .assertHasClickAction()
}

The useUnmergedTree = true set on the inner Text finder propagates to onParent() automatically — no need to repeat it.

Pattern: onAncestors() for the chain to root

kotlin
@Test
fun confirmButton_isInsideDialog() {
    rule.setContent { ConfirmDialog() }

    rule.onNodeWithTag(ConfirmButtonTag)
        .onAncestors()
        .filterToOne(isDialog())
        .assertExists()
}

PREFERRED: rule.onNode(hasTestTag(ConfirmButtonTag) and hasAnyAncestor(isDialog())).assertExists() — same intent in one matcher. See ../composing-semantics-matchers/SKILL.md.

Pattern: LazyColumn — only on-screen children appear

kotlin
// WRONG
rule.onNodeWithTag(ListTag).onChildren().assertCountEquals(1000)
// WRONG because: onChildren() returns a snapshot of currently composed children. A LazyColumn
// only composes the visible viewport plus a small prefetch buffer, so the count is window-sized,
// not data-set-sized.
kotlin
// RIGHT — assert the visible count, OR scroll first then assert by tag
rule.onNodeWithTag(ListTag).performScrollToIndex(999)
rule.onNodeWithTag("$ItemTagPrefix${999}").assertIsDisplayed()

See ../../patterns/testing-lazy-lists/SKILL.md for the full LazyColumn workflow.

Pattern: onSibling() to assert "the row's other half"

kotlin
@Test
fun checkbox_label_isPresent() {
    rule.setContent {
        Row {
            Checkbox(checked = true, onCheckedChange = {},
                modifier = Modifier.testTag("agreeBox"))
            Text("I agree", modifier = Modifier.testTag("agreeLabel"))
        }
    }

    rule.onNodeWithTag("agreeBox")
        .onSibling()
        .assertTextEquals("I agree")
}

onSibling() requires exactly one sibling (Selectors.kt:114-125). For multiple siblings use onSiblings() plus filterToOne(...).

Mandatory rules

  • MUST prefer adding a Modifier.testTag(...) to the target node over a multi-step traversal chain. Skydoves hot take #1.
  • MUST use filterToOne(matcher) over filter(matcher).onFirst() when the contract is "exactly one match"; the former throws on >1, the latter silently picks the first.
  • MUST NOT assume onChildren() returns the full data set for a LazyColumn / LazyRow — it returns the currently-composed snapshot only. Scroll first with performScrollToIndex / performScrollToKey.
  • MUST NOT look up a singular onAncestor — only onAncestors() plural exists. Use onParent() for the immediate parent or hasParent(matcher) / hasAnyAncestor(matcher) as predicates.
  • MUST remember the useUnmergedTree flag is sticky across onChild/onParent/filter. Skydoves hot take #2.
  • PREFERRED: for "anywhere above/below" relationships, replace traversal chains with hasAnyAncestor / hasAnyDescendant from ../composing-semantics-matchers/SKILL.md.

Verification

  • Every traversal chain has at most one navigation step, OR the deeper chain is justified by a comment naming the missing tag.
  • No onChildren()[i] chain spans a LazyColumn / LazyRow boundary without a preceding performScrollTo*.
  • No onAncestor (singular) usage — only onParent, onAncestors, or hasAnyAncestor.
  • filterToOne is used wherever the contract is "exactly one match"; filter().onFirst() is replaced.
  • useUnmergedTree = true on a chain root is intentional — there is a comment or matcher reason for it.
  • ./gradlew :app:connectedDebugAndroidTest or :app:testDebugUnitTest passes.

References

Frequently asked questions

What does the Traversing The Semantics Tree AI skill do?

Use this skill to navigate from one Compose semantics node to its relatives via `onParent`, `onChildren`, `onChild`, `onChildAt`, `onSibling`, `onSiblings`, `onAncestors`, plus the collection helpers `onFirst`, `onLast`, `filter`, `filterToOne`, and the `[index]` operator. Covers when to traverse vs when to add a stable `testTag`, the LazyColumn/LazyRow caveat (only currently composed children appear), the absence of a singular `onAncestor`, and the sticky `useUnmergedTree` flag across navigation. Use when the developer mentions `onChildren`, `onChild`, `onParent`, `onSiblings`, `onAncestor...

Why use Traversing The Semantics Tree on TypingMind?

Because you install it once and use it with any model. Traversing The Semantics Tree is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Traversing The Semantics Tree in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/skydoves/android-testing-skills/tree/main/compose/finders/traversing-the-semantics-tree. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Traversing The Semantics Tree?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Traversing The Semantics Tree?

As many as you like. As long as a model supports skills, you can use Traversing The Semantics Tree with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Traversing The Semantics Tree AI skill free?

Yes. It is published on GitHub by skydoves under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇