{ "SchemaVersion": "1", "Kind": "DirectoryIssues", "Slug": "scout-felixgeelhaar", "Name": "Scout", "CanonicalUrl": "https://askpod.ai/mcp/scout-felixgeelhaar/issues", "ServerUrl": "https://askpod.ai/mcp/scout-felixgeelhaar", "IssueTotal": 13, "Held": 13, "Issues": [ { "Title": "Stability + UX bugs surfaced during E2E driving of a Vue/Astro app", "Excerpt": "Feedback after using scout MCP (`scout 1.6.1`) to drive a full E2E walkthrough of a Vue + Astro + Go app (brotwerk). Most-impactful fix at top.\n\n## Stability bugs\n\n1. **Stale CDP node IDs after framework re-render.** `click`, `type`, `fill_form_semantic` cache CDP node refs from prior `annotated_screenshot` / `observe`. When Vue/React reconciles, refs invalidate but tools still use them → cryptic `cdp: error -32000: Could not find node with given id`. Fix: re-resolve selector to live nodeId at a", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/14", "PublishedAt": "2026-05-20T11:12:55.000Z", "State": "closed", "Comments": 1, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "Expose video recording (Playwright recordVideo) via MCP", "Excerpt": "## Problem\n\nScout exposes \\`start_recording\\` / \\`stop_recording\\` / \\`save_playbook\\` / \\`replay_playbook\\` — but those record an *action playbook* (JSON of clicks/types for replay), not a screen video.\n\nFor demo / docs / landing-page work, the value is the **video** (MP4 / WebM) that you can embed. Playwright already supports this natively via \\`browser.newContext({ recordVideo: { dir, size } })\\` — scout just doesn't expose it.\n\n## Workaround\n\nCall Playwright directly:\n\n\\`\\`\\`js\nconst ctx = a", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/4", "PublishedAt": "2026-05-02T15:02:18.000Z", "State": "closed", "Comments": 1, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "CDP connection stability + React compatibility issues", "Excerpt": "## Summary\n\nWhile using Scout extensively to automate and demo a React SPA dashboard ([go-teamhealthcheck](https://github.com/felixgeelhaar/go-teamhealthcheck)), I encountered several issues that made the experience unreliable. Filing them together since they came up in the same session.\n\n## Environment\n\n- Scout MCP server (via Claude Code)\n- Target: React 19 SPA with react-router-dom v7, served from Go backend on localhost\n- Both headless and visible modes tested\n\n## Issues\n\n### 1. Frequent CDP", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/1", "PublishedAt": "2026-03-26T19:12:37.000Z", "State": "closed", "Comments": 1, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "screenshot: add save_to_disk / output_path so captures return a file path instead of base64", "Excerpt": "## Problem\n\n`screenshot` and `annotated_screenshot` return a base64 data URL inline. For an agent driving scout on behalf of a user, that image lands in the **agent's** context — it renders for the model but never reaches the human. There is no way to hand the capture over.\n\nEncountered while running a scripted end-to-end walkthrough with screenshot milestones: every capture came back as base64, and the user simply saw nothing.\n\n## Current workarounds, both bad\n\n1. **Force an overflow.** Request", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/79", "PublishedAt": "2026-08-01T19:51:31.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "No viewport / device emulation: width params don't set the CSS viewport (responsive testing impossible)", "Excerpt": "## Summary\n\nThere is no way to set the browser **viewport size / emulate a device** through the MCP tools. Width parameters that *look* like they should control it (`screenshot.max_width`, `start_screen_recording.width/height`) only affect the **output image / capture frame**, not the page's CSS viewport. As a result, width-based `@media` breakpoints can't be exercised, so mobile/responsive layouts can't be verified.\n\n## Repro\n\n1. `configure { headless: true, fresh: true }` (also reproduces with", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/47", "PublishedAt": "2026-06-28T17:09:44.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "MCP: network capture returns zero requests after navigate() on a localhost SPA", "Excerpt": "## Summary\n\nWhen debugging a same-origin SPA served on `127.0.0.1`, the network-capture tools report **zero** captured requests — not even the document or static JS assets — despite the page clearly loading and issuing periodic `fetch()` polls. This makes it impossible to answer the most common debugging question (\"did this request fire, and what status did it return?\").\n\n## Environment\n\n- Scout MCP server (69-tool), driving headless Chromium\n- `configure` with `headless=true`, `allow_private_ip", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/42", "PublishedAt": "2026-06-09T09:13:18.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "Nice-to-haves: wait_for_navigation, dom_diff, extract_aria_violations", "Excerpt": "Three low-priority adds that would each save round-trips:\n\n## 1. `wait_for_navigation`\n`click(wait:true)` waits for full nav. SPA route changes (History API push) don't fire a real nav, so the option silently no-ops. Want a dedicated `wait_for_navigation({timeout, mode: \"full\" | \"spa\" | \"any\"})` that detects History API mutations + URL change + idle network.\n\n## 2. `dom_diff`\nCompare two `observe` snapshots and return only what changed:\n```\n{\n added: [...],\n removed: [...],\n changed_text: [{s", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/24", "PublishedAt": "2026-05-22T12:14:57.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "Add network_summary({since: \"last_action\"}) — collapse capture/check round-trip", "Excerpt": "## Problem\nToday: `enable_network_capture(patterns)` → action → `failed_requests` → maybe `console_errors`. Three calls for one question (\"did the click trigger a failed request?\").\n\n## Proposal\nSingle tool: `network_summary({since: \"last_action\" | \"last_navigation\" | \"session_start\"})`:\n```\n{\n total: 12,\n by_status: {\"2xx\": 10, \"4xx\": 1, \"5xx\": 1},\n failures: [\n {method: \"POST\", url: \"/api/auth/signup\", status: 500, body_snippet: \"...\"}\n ],\n pending: 0,\n duration_ms: 230\n}\n```\n\n## Why\n", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/23", "PublishedAt": "2026-05-22T12:14:37.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "top", "Extractor": "github_issue" }, { "Title": "type should report committed value after framework microtask flush", "Excerpt": "## Problem\n`type` returns the typed string immediately:\n```\n{selector: \"input[type=email]\", value: \"x@y.z\", action: \"typed\"}\n```\nBut on frameworks with two-way binding (Vue v-model, Svelte bind:value), the value may be transformed, sanitized, or never committed. Caller has no way to know without a separate `observe`.\n\n## Proposal\nMirror what `fill_form_semantic` already does:\n```\n{\n selector: \"input[type=email]\",\n typed: \"x@y.z\",\n committed: \"x@y.z\", // re-read after rAF + microtas", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/22", "PublishedAt": "2026-05-22T12:14:18.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "recent", "Extractor": "github_issue" }, { "Title": "annotated_screenshot: add stable node_handle alongside per-call label", "Excerpt": "## Problem\n`annotated_screenshot` warns that labels are per-call only — re-shuffled on any DOM mutation. Multi-step flows (annotate → click → annotate again) can't reliably reference the same node, forcing the caller to recompute selectors.\n\n## Proposal\nReturn a stable `node_handle` per element alongside the per-call label number:\n```\n{label: 7, node_handle: \"h_a3f2\", selector: \"button.mode-tab\", text: \"Mit Passwort\", ...}\n```\n- `node_handle` valid for as long as the node is alive (weak ref).\n- ", "SourceUrl": "https://github.com/klarlabs-studio/scout/issues/19", "PublishedAt": "2026-05-22T12:13:15.000Z", "State": "closed", "Comments": 0, "Reporter": "Maintainer", "Rank": "recent", "Extractor": "github_issue" }, { "Title": "Form submission silently fails: click_label on type=submit and dispatch_event(submit) don't fire @submit handlers", "Excerpt": "## Summary\n\nScout cannot drive form submission on a Vue 3 island that uses `@submit.prevent`. The handler never fires; no network request is dispatched; no console error surfaces. Two approaches both silently no-op:\n\n1. `click_label` on the `