@zenalexa/unicli 0.217.0 → 0.217.3

package/docs/operate/troubleshooting.md

@@ -0,0 +1,379 @@
+# Compute Troubleshooting
+
+`unicli compute` failures include `minimum_capability` and, where known, a
+structured `remedy`. Use this page when `unicli doctor compute` or a failed
+compute command points at one of these keys.
+
+## desktop-uia.binary_missing
+
+Cause: the Windows UIA sidecar binary is not installed or is not on the expected
+path.
+
+Remedy: run `unicli doctor compute --install`, then retry the compute command.
+
+Fallback: use CDP-backed Electron/browser control or CUA screenshot fallback if
+configured.
+
+## desktop-uia.startup_failed
+
+Cause: the UIA sidecar started and exited before responding.
+
+Remedy: run `UNICLI_TRACE=1 unicli doctor compute` and inspect antivirus,
+Defender, or SmartScreen quarantine.
+
+Fallback: retry after reinstalling the sidecar package.
+
+## desktop-uia.permission
+
+Cause: Windows denied the UIA operation.
+
+Remedy: run from an elevated terminal or install the sidecar with UIAccess.
+
+Fallback: use CDP for Electron/browser targets.
+
+## desktop-uia.no_element
+
+Cause: the saved ref no longer points at a live element.
+
+Remedy: run `unicli compute snapshot` again, then retry with a fresh ref.
+
+Fallback: target by app/window and take a new snapshot.
+
+## desktop-uia.not_invokable
+
+Cause: the target element does not expose the UIA Invoke pattern.
+
+Remedy: use `unicli compute type`, `unicli compute press`, or focus the control
+first and retry.
+
+Fallback: use CUA when no structured pattern is exposed.
+
+## desktop-uia.timeout
+
+Cause: the UIA sidecar did not respond within the request timeout.
+
+Remedy: retry the command; the sidecar boundary is designed to restart after
+timeouts.
+
+Fallback: run `UNICLI_TRACE=1 unicli doctor compute` if the timeout repeats.
+
+## desktop-uia.sidecar_crashed
+
+Cause: the UIA sidecar process exited or closed its pipe while a request was in
+flight.
+
+Remedy: retry once. If it repeats, run
+`UNICLI_TRACE=1 unicli doctor compute`.
+
+Fallback: use CDP for Electron/browser targets while inspecting sidecar logs.
+
+## Windows UIA Native Scope
+
+Current Windows UIA support has two layers:
+
+- Top-level app/window inventory, snapshot, find, wait, observe, assert, focus,
+  invoke, type, scroll, screenshot, and direct launch helpers use native
+  Win32/UIA-adjacent calls such as `EnumWindows`, `SetForegroundWindow`,
+  `SendInput`, HWND/GDI capture, and PowerShell `Start-Process`.
+- Live descendant UIA traversal populates refs where the OS exposes a control
+  tree. Descendant invoke, value, and focus actions prefer UIA patterns before
+  bounded fallback paths; descendant screenshot crops the owning window bitmap.
+
+Stable top-level refs look like `desktop-uia:pid-1234:Window[0]`. When passed to
+wait, Uni-CLI polls the native top-level window inventory for a matching
+role/name/title/app/pid filter until timeout. Observe ranks the same top-level
+window refs by goal/title token overlap. Assert checks the same inventory for
+top-level title text and visible/appear/enabled state.
+
+## desktop-atspi.binary_missing
+
+Cause: the Linux AT-SPI sidecar binary is not installed or is not on the
+expected path.
+
+Remedy: run `unicli doctor compute --install`, then retry.
+
+Fallback: use CDP-backed Electron/browser control or CUA screenshot fallback if
+configured.
+
+## desktop-atspi.dbus_blocked
+
+Cause: the AT-SPI bus daemon is not reachable.
+
+Remedy: run `systemctl --user start at-spi-dbus-bus`.
+
+Fallback: restart the desktop session if the bus cannot be started.
+
+## desktop-atspi.no_a11y_attr
+
+Cause: the target app does not expose a usable AT-SPI tree.
+
+Remedy: enable accessibility support for the app. Electron apps may need an
+accessibility flag at launch.
+
+Fallback: use CDP for Electron/browser targets or CUA screenshot fallback.
+
+## desktop-atspi.atspi_apps
+
+Cause: the Linux top-level app inventory helper is missing or failed.
+
+Remedy: ensure the AT-SPI bus is running. Uni-CLI prefers `wmctrl -lG -p` for
+real X11 window ids and geometry, but can fall back to AT-SPI-only registry
+roots when `wmctrl` is missing or empty.
+
+Fallback: target an Electron/browser app through CDP, or use CUA fallback.
+
+## desktop-atspi.atspi_windows
+
+Cause: the Linux top-level window inventory helper is missing or failed.
+
+Remedy: install `wmctrl` when real X11 window ids are needed. On Wayland or
+minimal environments, verify the AT-SPI bus is running so Uni-CLI can use
+synthetic `atspi-root-N` windows from the accessibility registry.
+
+Fallback: use `compute snapshot` on another transport or CUA fallback.
+
+## desktop-atspi.wayland-input
+
+Cause: Wayland input fallback tools are missing.
+
+Remedy: install `wtype` for text and printable key dispatch. Install `ydotool`
+for supported modifier combos and Wayland scroll fallback.
+
+Fallback: use structured AT-SPI actions that do not need synthetic input.
+
+## desktop-atspi.x11-input
+
+Cause: X11 input fallback tools are missing.
+
+Remedy: install `xdotool` for text, key, and scroll dispatch.
+
+Fallback: use structured AT-SPI actions that do not need synthetic input.
+
+## desktop-atspi.atspi_screenshot
+
+Cause: no supported Linux screenshot helper is available, or the helper failed.
+
+Remedy: install ImageMagick `import` for X11 top-level window capture, or
+install `grim` for Wayland top-level bounds capture. Install
+`gnome-screenshot` / `grim` for display-server fallback capture.
+
+Fallback: use CDP for Electron/browser targets or CUA screenshot fallback.
+
+## desktop-atspi.invalid_input
+
+Cause: a Linux AT-SPI action received a `ref` or `stable` token that is not a
+`desktop-atspi:pid-<pid>:Window[<n>]` top-level window token.
+
+Remedy: run `unicli compute snapshot --format compact` or
+`unicli compute find --first` on Linux, then retry with the fresh ref emitted by
+the AT-SPI transport.
+
+Fallback: omit the ref to use the global helper path for text, scroll, or
+screenshot actions when targeting the active desktop surface is acceptable.
+
+## desktop-atspi.no_element
+
+Cause: the saved ref no longer points at a live element.
+
+Remedy: run `unicli compute snapshot` again, then retry with a fresh ref.
+
+Fallback: target by app/window and take a new snapshot.
+
+## Linux AT-SPI Native Scope
+
+Current Linux AT-SPI support has two layers:
+
+- Top-level app/window inventory, snapshot, find, wait, observe, assert, focus,
+  invoke, type, scroll, screenshot, and direct launch helpers use `wmctrl` when
+  available, and AT-SPI-only synthetic windows when `wmctrl` is missing or
+  empty. Synthetic windows use native AT-SPI focus rather than `wmctrl`.
+- Descendant AT-SPI traversal populates role/name/value/state/bounds refs where
+  the app exposes an accessibility tree. Descendant invoke, type, and focus
+  prefer native Action/Value/EditableText/Component proxies, then fall back to
+  bounded display-server helpers. Descendant scroll prefers native
+  `Component.scroll_to(...)`, then falls back to display-server wheel helpers.
+- Screenshot uses `import -window <id>` for real X11 top-level window refs,
+  `grim -g` for known Wayland/top-level bounds, and region capture for bounded
+  descendant refs.
+
+Stable top-level refs look like `desktop-atspi:pid-1234:Window[0]`. When passed
+to type, scroll, or screenshot actions, Uni-CLI uses native operations when
+available and activates the top-level window only for helper fallbacks. On X11
+with ImageMagick `import`, screenshot uses the resolved top-level window id
+directly. On Wayland with `grim` and known bounds, screenshot captures the
+reported rectangle. Assert checks top-level and descendant text/value/state
+where the tree exposes them. Observe reports `action: "scroll"` for scroll
+roles or scrollable states.
+
+## desktop-atspi.sidecar_crashed
+
+Cause: the AT-SPI sidecar process exited or closed its pipe while a request was
+in flight.
+
+Remedy: retry once. If it repeats, run
+`UNICLI_TRACE=1 unicli doctor compute`.
+
+Fallback: use CDP for Electron/browser targets while inspecting sidecar logs.
+
+## desktop-ax.permission
+
+Cause: macOS Accessibility is not granted to the app or terminal that launches
+Uni-CLI.
+
+Remedy: grant Accessibility in System Settings, then retry. Deeplink:
+`x-apple.systempreferences:com.apple.preference.security?Privacy_Accessibility`.
+
+Fallback: CDP can still control browser/Electron renderers that expose a debug
+port.
+
+## desktop-ax.screen-recording
+
+Cause: macOS Screen Recording is not granted. This is only needed for screenshot
+fallback.
+
+Remedy: grant Screen Recording in System Settings. Deeplink:
+`x-apple.systempreferences:com.apple.preference.security?Privacy_ScreenCapture`.
+
+Fallback: prefer AX snapshots when the target exposes a structured tree.
+
+## desktop-ax.binary_missing
+
+Cause: the Swift runtime required by the macOS AX helper is not available.
+
+Remedy: run `xcode-select --install`.
+
+Fallback: use CDP for browser/Electron targets.
+
+## cdp-browser.attach_failed
+
+Cause: the requested CDP port is not reachable.
+
+Remedy: check the port, or start the browser transport with
+`unicli browser start`.
+
+Fallback: use native AX/UIA/AT-SPI snapshots for non-browser app control.
+
+## cdp-browser.electron_running_without_debug_port
+
+Cause: the Electron app is running, but no remote debugging port is available.
+
+Remedy: relaunch the app with a debug port, for example
+`unicli compute launch <app> --debug-port 9229`.
+
+Fallback: use native AX/UIA/AT-SPI controls when the app exposes enough
+accessibility structure.
+
+## subprocess.launcher
+
+Cause: the host launcher required by `compute launch` is unavailable. Uni-CLI
+uses `/usr/bin/open` on macOS, PowerShell `Start-Process` on Windows, and
+`gtk-launch` on Linux.
+
+Remedy: restore the platform launcher. On Linux install GTK desktop utilities,
+for example `sudo apt-get install libgtk-3-bin`.
+
+Fallback: start the app manually, then use `compute attach`, `compute snapshot`,
+or the native desktop transport for follow-up actions.
+
+## cua.no_backend
+
+Cause: screenshot/VLM fallback is enabled in the cascade, but no backend API key
+is configured.
+
+Remedy: set a supported backend environment variable such as
+`ANTHROPIC_API_KEY`, `TRYCUA_API_KEY`, or `OPENAI_API_KEY`.
+
+Fallback: use structured AX/UIA/AT-SPI/CDP control where possible.
+
+## compute.compute_find.ref-store
+
+Cause: `compute find --first` could not find a matching ref in the current ref
+store.
+
+Remedy: run `unicli compute snapshot`, then retry the find with the new refs.
+
+Fallback: make the snapshot more specific with `--app`.
+
+## compute.step.element_off_screen
+
+Cause: the ref is valid, but the element bounds are outside the visible window
+or screen.
+
+Remedy: scroll the containing view into range, then take a fresh
+`unicli compute snapshot`.
+
+Fallback: use `--focus` only when the app cannot expose a scrollable structured
+container.
+
+## compute.step.window_minimized
+
+Cause: the target window is minimized or hidden, so the transport cannot act in
+background mode.
+
+Remedy: restore the window or retry the action with explicit focus.
+
+Fallback: use `unicli compute windows --app <name>` to choose a visible window.
+
+## compute.step.element_disabled
+
+Cause: the target element exists but is disabled.
+
+Remedy: wait for it to become enabled with
+`unicli compute wait --state enabled`.
+
+Fallback: snapshot the surrounding UI and act on the prerequisite control.
+
+## compute.step.ref_expired
+
+Cause: the saved ref came from an older snapshot and no longer maps to a live
+element.
+
+Remedy: run `unicli compute snapshot`, then retry with the new ref.
+
+Fallback: use `compute find --first` after the new snapshot to select the ref.
+
+## compute.step.sidecar_crashed
+
+Cause: the UIA or AT-SPI sidecar exited while a call was in flight.
+
+Remedy: retry once. If it repeats, run
+`UNICLI_TRACE=1 unicli doctor compute`.
+
+Fallback: use CDP for Electron/browser targets while inspecting sidecar logs.
+
+## compute.step.sidecar_busy
+
+Cause: the sidecar is already processing a request.
+
+Remedy: retry after the current call completes. Sidecar calls are serialized to
+keep ref state stable.
+
+Fallback: avoid concurrent writes to the same app/window.
+
+## compute.step.app_ambiguous
+
+Cause: multiple running apps or windows match the same app name.
+
+Remedy: run `unicli compute windows --app <name>` and retry with a more
+specific bundle id, process name, pid, or window id.
+
+Fallback: target a CDP port or exact bundle id when controlling Electron apps.
+
+## compute.step.focus_required
+
+Cause: the transport cannot complete the action in background mode.
+
+Remedy: retry with `--focus` only after confirming focus stealing is acceptable.
+
+Fallback: prefer a structured value/action path such as `compute type` or
+`compute press` before enabling focus.
+
+## compute.step.no-transport-available
+
+Cause: every transport in the compute cascade failed or was unavailable for the
+requested step.
+
+Remedy: run `unicli doctor compute` and fix the first failing host transport.
+
+Fallback: choose a narrower transport directly when possible, such as CDP for a
+debuggable Electron renderer.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@zenalexa/unicli",
-  "version": "0.217.0",
+  "version": "0.217.3",
   "description": "Agent execution substrate for web, apps, local tools, and system capabilities",
   "type": "module",
   "main": "dist/main.js",
@@ -9,6 +9,9 @@
   },
   "files": [
     "dist/",
+    "crates/",
+    "docs/operate/",
+    "docs/mcp/clients/",
     "src/adapters/",
     "src/hub/",
     "AGENTS.md"
@@ -127,6 +130,9 @@
     "build:agents": "tsx scripts/build-agents.ts",
     "test": "vitest run --project unit",
     "test:adapter": "vitest run --project adapter",
+    "test:perf": "vitest run --project perf",
+    "compute:smoke": "tsx scripts/compute-live-smoke.ts",
+    "coverage:compute-snapshot": "vitest run --project unit tests/unit/refs.test.ts tests/unit/snapshot-encoder.test.ts --coverage --coverage.enabled=true --coverage.provider=v8 --coverage.include=src/transport/refs.ts --coverage.include=src/transport/snapshot-encoder.ts --coverage.reporter=text --coverage.thresholds.lines=100 --coverage.thresholds.functions=100 --coverage.thresholds.branches=100 --coverage.thresholds.statements=100",
     "coverage:adapter-test": "tsx scripts/check-adapter-test-coverage.ts --threshold 50",
     "adapter:bootstrap": "tsx scripts/bootstrap-adapter-tests.ts",
     "adapter:health": "tsx scripts/adapter-health-probe.ts",
@@ -137,17 +143,20 @@
     "format:check": "prettier --check .",
     "lint:context": "bash scripts/lint-context.sh",
     "build:manifest": "tsx scripts/build-manifest.js",
+    "sidecars:build": "node scripts/build-sidecars.mjs",
+    "sidecars:publish": "node scripts/publish-sidecars.mjs",
     "changeset": "changeset",
     "changeset:version": "changeset version",
     "changeset:status": "changeset status --since=origin/main",
     "verify:changesets": "tsx scripts/verify-changesets.ts",
+    "verify:release-mainline": "tsx scripts/release-mainline-check.ts",
     "conformance": "tsx scripts/conformance-report.ts",
     "check:exports": "tsx scripts/check-exports-count.ts",
     "lint:adapters": "tsx src/main.ts lint src/adapters",
     "lint:schema-v2": "tsx scripts/lint-schema-v2.ts",
     "clean": "node -e \"require('node:fs').rmSync('dist',{recursive:true,force:true})\"",
     "validate:agents-size": "tsx scripts/validate-agents-size.ts",
-    "verify": "npm run format:check && npm run typecheck && npm run lint && npm run lint:context && npm run build:manifest && npm run lint:adapters && npm run lint:schema-v2 && npm run build && npm run test && npm run test:adapter && npm run coverage:adapter-test && npm run validate:agents-size && npm run stats:check && npm run conformance && npm run check:exports && npm run verify:changesets",
+    "verify": "npm run format:check && npm run typecheck && npm run lint && npm run lint:context && npm run build:manifest && npm run lint:adapters && npm run lint:schema-v2 && npm run build && npm run test && npm run test:adapter && npm run test:perf && npm run coverage:compute-snapshot && npm run coverage:adapter-test && npm run validate:agents-size && npm run stats:check && npm run conformance && npm run check:exports && npm run verify:changesets",
     "verify:full": "npm run verify && npm run refs:verify && npm run adapter:health",
     "verify:clean": "npm run clean && npm run verify",
     "docs:prepare": "tsx scripts/generate-catalog.ts && tsx scripts/generate-docs-agent-assets.ts",
@@ -157,10 +166,10 @@
     "docs:preview": "vitepress preview docs",
     "doctor": "tsx src/doctor.ts",
     "mcp": "tsx src/mcp/server.ts",
-    "release": "npm run build && tsx scripts/release.ts",
+    "release": "npm run verify:release-mainline && npm run build && tsx scripts/release.ts",
     "release:check": "tsx scripts/release-check.ts",
     "prepare": "lefthook install >/dev/null 2>&1 || true",
-    "preversion": "npm run verify",
+    "preversion": "npm run verify:release-mainline && npm run verify",
     "postversion": "npm run build && tsx scripts/release.ts",
     "bench": "tsx bench/report.ts",
     "bench:self-discovery": "tsx bench/self-discovery.ts",
@@ -209,20 +218,36 @@
     "ws": "^8.20.0",
     "zod": "^4.3.6"
   },
+  "optionalDependencies": {
+    "@zenalexa/unicli-atspi-linux-arm64": "0.218.0",
+    "@zenalexa/unicli-atspi-linux-x64": "0.218.0",
+    "@zenalexa/unicli-uia-win32-arm64": "0.218.0",
+    "@zenalexa/unicli-uia-win32-x64": "0.218.0"
+  },
   "devDependencies": {
     "@ai-sdk/openai-compatible": "^2.0.41",
+    "@base-ui/react": "^1.4.1",
     "@changesets/cli": "^2.31.0",
     "@chenglou/pretext": "^0.0.6",
+    "@hugeicons/core-free-icons": "^4.1.1",
+    "@hugeicons/react": "^1.1.6",
+    "@hugeicons/vue": "^1.0.5",
     "@types/js-yaml": "^4.0.9",
     "@types/node": "^25.6.0",
+    "@types/react": "^19.2.14",
+    "@types/react-dom": "^19.2.3",
     "@types/shell-quote": "^1.7.5",
     "@types/turndown": "^5.0.6",
     "@types/ws": "^8.18.1",
+    "@vitejs/plugin-react": "^6.0.1",
+    "@vitest/coverage-v8": "^3.2.4",
     "ai": "^6.0.168",
     "lefthook": "^2.1.6",
     "markdown-it-mathjax3": "^4.3.2",
     "oxlint": "^1.61.0",
     "prettier": "^3.8.3",
+    "react": "^19.2.5",
+    "react-dom": "^19.2.5",
     "shell-quote": "^1.8.3",
     "tsx": "^4.21.0",
     "typescript": "^6.0.3",

package/src/adapters/juejin/hot.test.ts

@@ -0,0 +1,25 @@
+import { describe, it } from "vitest";
+import {
+  runAdapterWithFixture,
+  expectAdapterShape,
+} from "../../../tests/adapter-runner.js";
+
+describe("juejin hot", () => {
+  it("returns rows with declared columns against fixture", async () => {
+    const { output } = await runAdapterWithFixture("juejin", "hot");
+    expectAdapterShape(output, {
+      columns: [
+        "rank",
+        "title",
+        "author",
+        "company",
+        "views",
+        "diggs",
+        "comments",
+        "hot_index",
+        "url",
+      ],
+      minItems: 1,
+    });
+  });
+});

package/src/adapters/juejin/hot.yaml

@@ -0,0 +1,52 @@
+site: juejin
+name: hot
+description: 掘金全站热门推荐 / Juejin all-categories hot feed
+domain: juejin.cn
+type: web-api
+strategy: public
+
+args:
+  limit:
+    type: int
+    default: 20
+    description: Number of items
+
+pipeline:
+  - fetch:
+      url: https://api.juejin.cn/recommend_api/v1/article/recommend_all_feed?aid=2608&uuid=&spider=0
+      method: POST
+      headers:
+        Content-Type: application/json
+        User-Agent: "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
+      body:
+        id_type: 2
+        client_type: 2608
+        sort_type: 200
+        cursor: "0"
+        limit: 30
+
+  - select: data
+
+  - filter: item.item_type === 2 && item.item_info && item.item_info.article_info
+
+  - map:
+      rank: ${{ index + 1 }}
+      title: ${{ item.item_info.article_info.title }}
+      author: ${{ item.item_info.author_user_info && item.item_info.author_user_info.user_name }}
+      company: ${{ item.item_info.author_user_info && item.item_info.author_user_info.company }}
+      views: ${{ item.item_info.article_info.view_count }}
+      diggs: ${{ item.item_info.article_info.digg_count }}
+      comments: ${{ item.item_info.article_info.comment_count }}
+      hot_index: ${{ item.item_info.article_info.hot_index }}
+      url: https://juejin.cn/post/${{ item.item_info.article_info.article_id }}
+
+  - limit: ${{ args.limit }}
+
+columns: [rank, title, author, company, views, diggs, comments, hot_index, url]
+
+capabilities: ["http.fetch"]
+minimum_capability: http.fetch
+trust: public
+confidentiality: public
+quarantine: false
+schema_version: v2

package/src/adapters/juejin/search.test.ts

@@ -0,0 +1,27 @@
+import { describe, it } from "vitest";
+import {
+  runAdapterWithFixture,
+  expectAdapterShape,
+} from "../../../tests/adapter-runner.js";
+
+describe("juejin search", () => {
+  it("returns rows with declared columns against fixture", async () => {
+    const { output } = await runAdapterWithFixture("juejin", "search", {
+      args: { query: "Anthropic" },
+    });
+    expectAdapterShape(output, {
+      columns: [
+        "rank",
+        "title",
+        "excerpt",
+        "author",
+        "company",
+        "views",
+        "diggs",
+        "comments",
+        "url",
+      ],
+      minItems: 1,
+    });
+  });
+});

package/src/adapters/juejin/search.yaml

@@ -0,0 +1,58 @@
+site: juejin
+name: search
+description: 掘金全文搜索 / Juejin full-text search
+domain: juejin.cn
+type: web-api
+strategy: public
+
+args:
+  query:
+    required: true
+    positional: true
+    description: Search keyword
+  limit:
+    type: int
+    default: 20
+    description: Number of results
+
+pipeline:
+  - fetch:
+      url: https://api.juejin.cn/search_api/v1/search
+      params:
+        aid: "2608"
+        uuid: ""
+        spider: "0"
+        query: "${{ args.query }}"
+        id_type: "0"
+        cursor: "0"
+        limit: "${{ args.limit }}"
+        search_type: "0"
+        sort_type: "0"
+      headers:
+        User-Agent: "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
+
+  - select: data
+
+  - filter: item.result_type === 2 && item.result_model && item.result_model.article_info
+
+  - map:
+      rank: ${{ index + 1 }}
+      title: ${{ item.result_model.article_info.title }}
+      excerpt: ${{ String(item.result_model.article_info.brief_content || '').slice(0, 150) }}
+      author: ${{ item.result_model.author_user_info && item.result_model.author_user_info.user_name }}
+      company: ${{ item.result_model.author_user_info && item.result_model.author_user_info.company }}
+      views: ${{ item.result_model.article_info.view_count }}
+      diggs: ${{ item.result_model.article_info.digg_count }}
+      comments: ${{ item.result_model.article_info.comment_count }}
+      url: https://juejin.cn/post/${{ item.result_model.article_info.article_id }}
+
+  - limit: ${{ args.limit }}
+
+columns: [rank, title, excerpt, author, company, views, diggs, comments, url]
+
+capabilities: ["http.fetch"]
+minimum_capability: http.fetch
+trust: public
+confidentiality: public
+quarantine: false
+schema_version: v2

package/src/adapters/leetcode/discuss-search.test.ts

@@ -0,0 +1,29 @@
+import { describe, it } from "vitest";
+import {
+  runAdapterWithFixture,
+  expectAdapterShape,
+} from "../../../tests/adapter-runner.js";
+
+describe("leetcode discuss-search", () => {
+  it("returns rows with declared columns against fixture", async () => {
+    const { output } = await runAdapterWithFixture(
+      "leetcode",
+      "discuss-search",
+      {
+        args: { query: "Anthropic" },
+      },
+    );
+    expectAdapterShape(output, {
+      columns: [
+        "rank",
+        "title",
+        "summary",
+        "author",
+        "views",
+        "created",
+        "url",
+      ],
+      minItems: 1,
+    });
+  });
+});